Use of a Brief Food Frequency Questionnaire for Estimating Daily ...

RESEARCHResearch and Professional BriefsUse of a Brief Food Frequency Questionnaire forEstimating Daily Number of Servings of Fruitsand Vegetables in a Minority AdolescentPopulationJENNIFER DI NOIA, PhD; ISOBEL R. CONTENTO, PhD, CDNJ. Di Noia is an assistant professor, Department of Sociology,William Paterson University, Wayne, NJ; at thetime of the study, she was a research scientist at IntersystemsIncorporated, Parsippany, NJ. I. Contento isMary Swartz Rose Professor of Nutrition and Education,Teachers College, Columbia University, New York,NY.Address correspondence to: Jennifer Di Noia, PhD,Department of Sociology, William Paterson University,300 Pompton Road, Wayne, NJ 07470. E-mail: dinoiaj@wpunj.edu.Manuscript accepted: April 6, 2009.Copyright © 2009 by the American DieteticAssociation.0002-8223/09/10910-0014$36.00/0doi: 10.1016/j.jada.2009.07.002ABSTRACTThe validity of the 5 A Day for Better Health Programfood frequency questionnaire (5 A Day FFQ) for estimatingfruit and vegetable consumption was examined in asample of 156 African-American adolescents aged 10 to14 years. To determine validity, the correlation between 5A Day FFQ fruit, juice, and vegetable intake and 3-dayintake measured by direct observation was assessed. Correlationswere calculated separately by food type (ie,fruits, juices, and vegetables) and sex and age to determinewhether the accuracy of youths’ recording differedbased on these factors. Paired-samples t tests were usedto test for differences between 5 A Day FFQ and observedintake estimates. The ability of the 5 A Day FFQ tocorrectly classify youths according to intake level (ie,intake of five or more daily servings) was also examined.The 5 A Day FFQ intake was significantly correlated withobserved intake (r0.39; P0.01). Correlations wereweaker by food type (r0.15 to r0.28) and did not differbased on youths’ sex or age. Mean 5 A Day FFQ intake(6.746.00 servings) was considerably higher than meanobserved intake (5.411.51 servings), and this was aresult of the overestimation of vegetable intake. Sensitivity,specificity, and positive and negative predictive valuesof the 5 A Day FFQ were 67.1%, 68.6%, 63.5%, and71.9%, respectively. Findings suggest that the 5ADayFFQ may be more useful as a screening tool for identifyingAfrican-American adolescents most in need of interventionthan for estimating youths’ mean intake in dietaryintervention programs.J Am Diet Assoc. 2009;109:1785-1789.Current dietary guidelines recommend consumptionof between 5 and 13 daily servings of fruits andvegetables (1), yet average American intakes arewell below recommended levels (2). Differences betweenrecommended and actual intake are more pronouncedamong African Americans (2), underscoring the need fordietary interventions to increase intakes in this population.The choice of assessment method for trackingchanges in intake is a key consideration in the design ofsuch programs.Food frequency questionnaires (FFQs) are commonlyused for this purpose (3). Data collection and processingcosts and respondent burden are lower for FFQs than forother intake assessment methods, such as multiple foodrecords and 24-hour recalls, features that make themattractive to program planners (4). Although they can beof considerable length, shorter questionnaires for measuringintakes of specific nutrients or food groups havebeen developed (5-8).The 5 A Day for Better Health Program FFQ (5 A DayFFQ) is a seven-item measure of fruit and vegetableconsumption (6). Correlations between 5 A Day FFQ intakeand intake measured by full-length FFQs, foodrecords, and 24-hour recalls range from 0.50 to 0.52, 0.19to 0.51, and 0.50 to 0.58, respectively (5,6,9-13). Relativeto these reference methods, the 5 A Day FFQ has tendedto underestimate intake.Validation studies of the 5 A Day FFQ have been conductedprimarily with adults (5,6,9-12). The purpose ofthis study was to examine the validity of the 5 A Day FFQfor estimating fruit and vegetable consumption amongAfrican-American adolescents. In a cross-sectional design,the correlation between 5 A Day FFQ intake and3-day intake measured by direct observation was assessed.A significant and moderate correlation (ie, 0.50by Cohen’s conventions) was considered evidence of validity(14). Correlations were calculated separately byfood type (ie, fruits, juices, and vegetables) and by sex andage to determine whether the accuracy of youths’ recordingdiffered based on these factors. The ability of the 5 ADay FFQ to correctly classify youths according to intakelevel (ie, intake of five or more daily servings) was alsoexamined. Objectives for this study were to validate the 5© 2009 by the American Dietetic Association Journal of the AMERICAN DIETETIC ASSOCIATION 1785

A Day FFQ as an assessment of fruit and vegetable consumptionamong African-American adolescents and determinethe effect of sex and age on the validity of the 5A Day FFQ.METHODSData were provided by African-American adolescents enrolledin a measurement validation study described elsewhere(15). Youths were recruited through summer daycamp programs offered at youth services agencies in NewYork City. One hundred eighty one youths between theages of 10 and 14 years were enrolled. Because observationaldata used to validate the 5 A Day FFQ were collectedduring meals served during 3 consecutive days (asdescribed here later), evaluation of the 5 A Day FFQ wasrestricted to youths who were present at all meals andthus for whom complete observational data were available(n156). Study procedures were approved by theInstitutional Review Board of Intersystems Incorporated.Prior to their study involvement, all youths providedwritten assent and informed written consent was obtainedfrom a parent or guardian.Youths were served breakfast, lunch, and dinner atcollaborating sites during 3 consecutive days. At eachmeal, youths were given a tray with one serving each (asdefined by 5 A Day criteria) of fruits, juices, and vegetablesand were offered a variety of main course options(16). Between meals, youths participated in recreationalactivities. They were not offered fruits and vegetables tosnack on at these other times.Trained staff present at meals observed the amounts offruit, juice, and vegetables youths left after eating byusing the plate-waste-by-visual-estimate method (17).They recorded the portion consumed of each serving on aform that contained the following response options: zero,one fourth, one half, three fourths, and one. Becausemeals were served in a large-group format, each staffmember observed different groups of up to 10 youthseach; they did not concurrently observe the same youthsat each meal.After dinner on the third day, youths were administeredthe 5 A Day FFQ, a nonquantitative measure thatqueries the number of times respondents ate or drank thefollowing items during the previous month: 100% orangejuice or grapefruit juice; other 100% fruit juices; greensalad; french fries or fried potatoes; baked, boiled, ormashed potatoes; other vegetables; and fruit. The specificwording of items is shown in the Figure. Because thereference periods for the measures differed (ie, the precedingmonth vs the preceding 3 days, respectively), 5 ADay FFQ and observed intake estimates were convertedto a common metric. Youths’ 5 A Day FFQ responses weretransformed to a daily equivalent and summed to providea composite measure of servings per day. Comparableestimates of observed intake were computed by determiningthe number of servings youths were observed eatingon each day and averaging these amounts across days.5 A Day FFQ and observed intake estimates by food type(ie, fruits, juices, and vegetables) were also calculated.The 5 A Day FFQ item for measuring intake of frenchfries and fried potatoes was excluded from scoring proceduresbecause these foods are high in fat and are notrecommended in the 5 A Day for Better Health ProgramIn the past month, how often did you drink or eat:100% orange juice or grapefruit juice?Other 100% fruit juices, NOT COUNTING fruit drinks?Green salad (with or without other vegetables)?French fries or fried potatoes?Baked, boiled, or mashed potatoes?Response options are: never, one to three times/month, one totwo times/week, three to four times/week, five to six times/week,one time/day, two times/day, three times/day, four times/day andfive or more times/day.In the past month, about how many servings of vegetables didyou eat, NOT COUNTING potatoes and salad?In the past month, about how many servings of fruit did you eat,NOT COUNTING juices?Response options are: none, one to three/month, one to two/week,three to four/week, five to six/week, one/day, two/day, three/day,four/day, and five or more/day.Figure. Questions asked on the 5 A Day for Better Health Program foodfrequency questionnaire.(18). French fries and fried potatoes were included in the5 A Day FFQ to help respondents distinguish them fromthe nonfried potatoes the measure was intended to count.The item for measuring these foods was specifically designedto identify and then omit them from intake estimatesof recommended foods and juices (19). French friesand fried potatoes were not included in meals provided toyouths; thus, they were also not reflected in observedintake estimates.Pearson correlations were calculated to provide an indexof relation between 5 A Day FFQ and observed intakeand intake by food type. To examine whether and how thecorrelations differed among the sex and age groups studied,youths were stratified by sex and age and separatecorrelations were calculated for males and females andfor younger (ie, aged 10 to 11 years) and older (ie, aged 12to 14 years) youths. To determine whether pairs of correlations(eg, males vs females) were statistically different,the Fisher transformation method was used (20). Pairedsamples t tests were used to test for differences between5 A Day FFQ and observed intake estimates.To examine the ability of the 5 A Day FFQ to correctlyclassify youths according to intake level, 5 A Day FFQand observed intake were dichotomized (fewer than fivedaily servings, five or more daily servings). The sensitivity,specificity, and positive and negative predictive valuesof the 5 A Day FFQ were examined using thesedichotomized scores. All analyses were conducted usingSPSS software (version 12.0.1 for Windows, 2003, SPSSInc, Chicago, IL).RESULTS AND DISCUSSIONParticipants had a mean age of 11.89 (1.24) years andwere 55% female. Although 5 A Day FFQ intake wassignificantly correlated with observed intake (r0.39;P0.01), youths significantly overestimated their intakeusing the 5 A Day FFQ (t 155 2.97; P0.01) (Table).1786 October 2009 Volume 109 Number 10

Table. Comparison of daily servings of fruits, juices, and vegetables and servings by food type estimated by the 5 A Day for Better HealthProgram food frequency questionnaire and observed intakeIntake measure n a (meanSD c )5 A Day FFQ bRangeObservedintake(meanSD)Range5 A Day FFQminus observedintake(differenceSE d )Pvalue eTotal intakeAll cases 156 6.746.00 29.80 5.411.51 8.25 1.330.45 0.01 0.39**Male 70 6.305.05 19.93 5.301.54 6.67 1.000.57 0.08 0.34**Female 86 7.106.69 29.80 5.511.50 8.17 1.590.67 0.01 0.43**10-11 years 64 6.416.37 28.80 5.591.49 8.08 0.820.75 0.28 0.36**12-14 years 92 6.975.76 29.80 5.291.52 6.67 1.680.55 0.01 0.43**Fruit intakeAll cases 147 1.591.80 5.00 1.610.72 3.33 0.020.15 0.91 0.24**Male 64 1.311.59 5.00 1.490.78 2.92 0.180.20 0.37 0.32*Female 83 1.811.93 5.00 1.700.66 3.25 0.110.21 0.60 0.23*10-11 years 61 1.291.70 5.00 1.560.69 3.17 0.270.75 0.23 0.2012-14 years 86 1.811.86 5.00 1.650.75 2.92 0.160.20 0.42 0.25*Juice intakeAll cases 156 2.922.89 10.00 2.610.49 3.08 0.310.23 0.18 0.15Male 70 3.102.88 10.00 2.730.35 1.92 0.370.35 0.28 0.02Female 86 2.772.92 10.00 2.520.57 2.83 0.250.31 0.41 0.22*10-11 years 64 2.962.97 10.00 2.690.43 2.42 0.270.37 0.47 0.1012-14 years 92 2.892.86 10.00 2.560.52 2.83 0.330.29 0.25 0.18Vegetable intakeAll cases 156 2.322.95 15.00 1.260.68 4.67 1.060.23 0.01 0.28**Male 70 2.012.47 10.21 1.210.67 2.67 0.800.29 0.01 0.16Female 86 2.571.30 15.00 1.300.68 4.58 1.270.34 0.01 0.34**10-11 years 64 2.223.07 15.00 1.370.70 4.33 0.850.37 0.03 0.26*12-14 years 92 2.392.88 15.00 1.190.65 2.67 1.200.29 0.01 0.30**a For the fruit category, the sample size was reduced due to the omission of cases with missing data (n9).b 5 A Day FFQ5 A Day for Better Health Program food frequency questionnaire.c SDstandard deviation.d SEstandard error.e Values shown indicate the level of significance for paired-samples t tests comparing 5 A Day FFQ and observed intake.*Pearson correlation between 5 A Day FFQ and observed intake significant at P0.05.**Pearson correlation between 5 A Day FFQ and observed intake significant at P0.01.Correlation between5 A Day FFQ andobserved intakeThe correlations were weaker by food type and weresignificant for fruits (r0.24; P0.01) and vegetables(r0.28; P0.01), but not juices (r0.15; P0.05). 5 ADay FFQ intake of fruits and juices did not differ fromobserved fruit and juice intake. However, 5 A Day FFQintake of vegetables was significantly higher than observedvegetable intake (t 155 4.65; P0.01). Correlationsbetween 5 A Day FFQ and observed intake andintake by food type did not differ based on youths’ sex orage.Sensitivity, specificity, and positive and negative predictivevalues of the 5 A Day FFQ were 67.1%, 68.6%,63.5%, and 71.9%, respectively. This means that 67.1% ofyouths with observed intake of fewer than five daily servingswere classified by the 5 A Day FFQ as consumingfewer than five daily servings, and 68.6% of youths withobserved intake of five or more daily servings were classifiedby the 5 A Day FFQ as consuming five or more dailyservings. Moreover, if the 5 A Day FFQ determined thatintake was fewer than five daily servings, it was correct63.5% of the time, and if it determined that intake wasfive or more daily servings, it was correct 71.9% of thetime. These performance indicators suggest that the 5 ADay FFQ is useful for confirming intake levels (as indicatedby its positive and negative predictive value) and itwill correctly classify sizable and similar proportions ofyouths with intakes below and at or above five dailyservings (as indicated by its sensitivity and specificity,respectively).This study examined the validity of the 5 A Day FFQfor estimating fruit and vegetable consumption amongAfrican-American adolescents. The correlation between 5A Day FFQ and observed intake was below the a prioricriterion, an indication that the 5 A Day FFQ was not avalid measure of intake in this population. Noteworthy isthe heterogeneity of responses to 5 A Day FFQ items asindicated by the large range of values for 5 A Day FFQintake estimates. Although the reference period for the 5A Day FFQ is the preceding month, response optionsquery the number of times per month, week, or day respondentsconsumed the included items. The large variationfound may be an indication that youths had difficultyquantifying their monthly intake in daily andOctober 2009 ● Journal of the AMERICAN DIETETIC ASSOCIATION 1787

weekly equivalents, a factor that may account for the poorvalidity of the measure.The overestimation found was consistent with findingsfrom a validation study of the 5 A Day FFQ in a sample ofchildren (13). In that study, youths overestimated theirintake regardless of food type, whereas in this study,youths overestimated their vegetable intake only. A numberof factors may explain this pattern.Observational data revealed that youths’ vegetable intakewas lower than their fruit and juice intake. Youths’tendency to consume more fruit (mostly through fruitjuice) than vegetables is consistent with intake patternsfound among African Americans in other research (21).Because youths consumed fruits and juices more oftenthan vegetables, they may have been recalled withgreater accuracy.Alternatively, the overestimation may be an artifact ofthe greater number of items used to measure vegetableintake. The 5 A Day FFQ uses three items to measurevegetable intake, two items to measure juice intake, andone item to measure fruit intake. In other research, reportedservings of fruits, juices, and vegetables increasedin direct proportion to the number of questions for measuringthese foods (22).Possibly, youths were influenced by social approval andsocial desirability biases, the tendencies of individuals torespond in a manner consistent with expected norms (23).Youths may have been aware that vegetables are foodsthey should be in the habit of eating. This awareness,combined with their low vegetable intake, may have encouragedthem to report consuming vegetables more oftenthan was the case.The small and self-selected sample limits the generalizabilityof findings. The time periods referenced by themeasures used differed. Youths may have had difficultytranslating servings consumed during 3 days to servingsconsumed in a month. Results may therefore underestimatethe true validity of the 5 A Day FFQ. Respondentsdid not have the option of selecting from among a varietyof fruits and vegetables at meals and they were not giventhese foods to snack on between meals. There was alsothe possibility that youths consumed additional fruitsand vegetables before and after camp. Thus, estimates ofobserved intake may not reflect youths’ true eating behavior.Previous research has shown that FFQs are susceptibleto seasonal reporting bias (24,25). Because thisstudy was conducted in the summer, results may be differentthan they would be at other times of the year.This study used observational data to validate the 5 ADay FFQ, a criterion measure superior to self-report measures(26). This is the first study to examine the validityof the 5 A Day FFQ among African-American adolescents.Findings advance understanding of the validity of thistool for estimating fruit and vegetable consumption inthis population.CONCLUSIONSThe poor validity of the 5 A Day FFQ suggests that it isless than optimal for program evaluation purposes. The 5A Day FFQ may overestimate the true intervention effector fail to find an effect on vegetable intake. Investigatorsare therefore encouraged to administer measures ofknown validity to confirm study findings based on the 5 ADay FFQ alone. The moderate sensitivity, specificity, andpositive and negative predictive value of the 5 A Day FFQsuggest that it may be more useful as a screening tool foridentifying African-American youths most in need of intervention.The decision to use the 5 A Day FFQ forscreening purposes should be based on tolerance for somemisclassification.STATEMENT OF POTENTIAL CONFLICT OF INTEREST:No potential conflict of interest was reported by the authors.FUNDING/SUPPORT: The work under considerationwas supported by funding from the National Cancer Institute(CA 096162).References1. US Department of Health and Human Services, US Department ofAgriculture. Dietary Guidelines for Americans, 2005. 6th ed. Washington,DC: US Government Printing Office; January 2005.2. Casagrande SS, Wang Y, Anderson C, Gary TL. Have Americansincreased their fruit and vegetable intake? Am J Prev Med. 2007;32:1-7.3. Hunt MK, Stoddard AM, Peterson K, Sorenson G, Herbert JR, CohenN. Comparison of dietary assessment measures in the Treatwell 5 ADay worksite study. J Am Diet Assoc. 1998;98:1021-1023.4. Thompson FE, Byers T. Dietary assessment resource manual. J Nutr.1994;124(suppl 11):2245-2317.5. Warneke CL, Davis M, De Moor C, Baranowski T. A 7-item versus31-item food frequency questionnaire for measuring fruit, juice, andvegetable intake among a predominantly African-American population.J Am Diet Assoc. 2001;101:774-779.6. Thompson FE, Kipnis V, Subar AF, Krebs-Smith SM, Kahle LL,Midthune D, Potischman N, Schatzkin A. Evaluation of 2 brief instrumentsand a food-frequency questionnaire to estimate daily number ofservings of fruits and vegetables. Am J Clin Nutr. 2000;71:1503-1510.7. Thompson FE, Subar AF, Smith AF, Midthune D, Radimer KL, KahleLL, Kipnis V. Fruit and vegetable assessment: Performance of 2 newshort instruments and a food frequency questionnaire. J Am DietAssoc. 2002;102:1764-1772.8. Block G, Gillespie C, Rosenbaum EH, Jenson C. Rapid food screenerto assess fat and fruit and vegetable intake. Am J Prev Med. 2000;18:284-288.9. Campbell MK, Demark-Wahnefried W, Symons M, Kalsbeek WD,Dodds J, Cowan A, Jackson B, Mostinger B, Hoben K, Lashley J,Demissie S, McClelland JW. Fruit and vegetable consumption andprevention of cancer: The Black Churches United for Better Healthproject. Am J Public Health. 1999;89:1390-1396.10. Snyder DC, Sloane R, Lobach D, Lipkus I, Clipp E, Kraus WE,Demark-Wahnefried W. Agreement between a brief mailed screenerand an in-depth telephone survey: Observations from the Fresh Startstudy. J Am Diet Assoc. 2004;104:1593-1596.11. Kristal AR, Vizenor NC, Patterson RE, Neuhouser ML, Shattuck AL,McLerran D. Precision and bias of food frequency-based measures offruit and vegetable intakes. Cancer Epidemiol Biomarkers Prev. 2000;9:939-944.12. Plesko M, Cotugna N, Aljadir L. Usefulness of a brief fruit andvegetable FFQ in a college population. Am J Health Behav. 2000;24:201-208.13. Baranowski T, Smith M, Baranowski J, Wang DT, Doyle C, Lin LS,Hearn MD, Resnicow K. Low validity of a seven-item fruit and vegetablefood frequency questionnaire among third-grade students. JAmDiet Assoc. 1997;97:66-68.14. Cohen J. Statistical Power Analysis for the Behavioral Sciences. 2nded. Hillsdale, NJ: Lawrence Erlbaum; 1988.15. Di Noia J, Contento IR. Criterion validity and user acceptability of aCD-ROM-mediated food record for measuring fruit and vegetableconsumption among black adolescents. Public Health Nutr. 2009;12:3-11.16. Eat 5 Fruits and Vegetables a Day. Washington, DC: National CancerInstitute, National Institutes of Health; 1999. NIH Publication 99-3201.17. US Department of Agriculture, Economic Research Service. PlateWaste in School Nutrition Programs: Final Report to Congress. http://www.ers.usda.gov/publications/efan02009/efan02009.pdf. AccessedApril 12, 2006.1788 October 2009 Volume 109 Number 10

18. Thompson B, Demark-Wahnefried W, Taylor G, McClelland JW, StablesG, Havas S, Feng Z, Topor M, Heimendinger J, Reynolds KD,Cohen N. Baseline fruit and vegetable intake among adults in seven5 A Day study centers located in diverse geographic areas. J Am DietAssoc. 1999;99:1241-1248.19. Marcus AC, Heimendinger J, Wolfe P, Rimer BK, Morra M, Cox, D,Lang PJ, Stengle W, Van Herle MP, Wagner D, Fairclough D, HamiltonL. Increasing fruit and vegetable consumption among callers tothe CIS: Results from a randomized trial. Prev Med. 1998;27:S16-S28.20. Hays WL. Statistics. 5th ed. Chicago, IL: Holt, Reinhart, and Winston;1994.21. Subar AF, Heimendinger J, Patterson BH, Krebs-Smith SM, PivonkaE, Kessler R. Fruit and vegetable intake in the United States: Thebaseline survey of the Five A Day for Better Health Program. Am JHealth Promot. 1995;9:352-360.22. Krebs-Smith SM, Heimendinger J, Subar AF, Patterson BH, PivonkaE. Using food frequency questionnaires to estimate fruit and vegetableintake: Association between the number of questions and totalintake. J Nutr Educ. 1995;27:80-85.23. Miller TM, Abdel-Maksoud MF, Crane LA, Marcus AC, Byers TE.Effects of social approval bias on self-reported fruit and vegetableconsumption: A randomized controlled trial. Nutr J. 2008;7:118-124.24. Subar AF, Frey CM, Harlan LC, Kahle L. Differences in reported foodfrequency by season of questionnaire administration: The 1987 NationalHealth Interview Survey. Epidemiology. 1994;5:226-233.25. Fowke JH, Schlundt D, Gong Y, Jin F, Xia-Ou S, Wen W, Da-Ke L,Yu-Tang G, Zheng W. Impact of season of food frequency questionnaireadministration on dietary reporting. Ann Epidemiol. 2004;14:778-785.26. Resnicow K, Odom E, Wang T, Dudley WN, Mitchell D, Vaughan R,Jackson A, Baranowski T. Validation of three food frequency questionnairesand 24-hour recalls with serum carotenoid levels in asample of African-American adults. Am J Epidemiol. 2000;152:1072-1080.October 2009 ● Journal of the AMERICAN DIETETIC ASSOCIATION 1789