Upload
damon-brown
View
215
Download
1
Embed Size (px)
Citation preview
Fobns
cciKeo“otsdiat
iNutcfhc
ouyscDPtega
ta
©
practice applicationsBEYOND THE HEADLINES
Do Food Frequency Questionnaires
Have Too Many Limitations?BsNLsd
OTtttc2tp
st1DwmutDevnc‘sibipvuTppf
rwmhtMt
trl
LTabvtqqhsfgsawl
HtFlmqtcdcpppcpm
cffpebfpFmiw
ood frequency questionnaires, orFFQs, are designed to measure aperson’s usual dietary intake
ver a defined period of time (1). Usedy researchers in some form forearly 50 years, FFQs have been con-idered a useful, if broad, tool.As of late, however, the device has
ome under some scrutiny. In a De-ember 2005 editorial for Cancer Ep-demiology Biomarkers Prevention,ristal and colleagues thought thatpidemiologists needed to face theverwhelming limitations of the FFQ.We are now facing a crisis: hundredsf millions of dollars and many scien-ists’ careers have been invested intudies using only FFQs to measureiet, but it is possible that these stud-es have not been, and will not be,ble to answer many if not most ques-ions about diet and cancer risk” (2).
A month later, the January 2006ssue of the European journal Foodavigator claimed that FFQs can bep to 50% inaccurate. “The reality ishat the FFQ is a fragile basis for anyonclusion, and yet it is currently theoundation for most of the food andealth stories that consumers re-eive” (3).Professionals admit the limitations
f FFQs, but say that it is a matter ofsing them well. “I like FFQs becauseou can get data on a lot of people,”ays Catherine Champagne, PhD, RD,hief of Nutritional Epidemiology/ietary Assessment and Counseling atennington Biomedical Research Cen-er, Baton Rouge, LA. “It gives a gen-ral view as opposed to specificity on aiven day. It has value. And it is quicknd easy to give.”“Regarding the criticism, it is naive
o think that there is one perfectssessment method,” says Carol
This article was written byDamon Brown, a freelancewriter in Long Beach, CA.
Mdoi: 10.1016/j.jada.2006.07.020
2006 by the American Dietetic Associat
oushey, PhD, RD, associate profes-or in the Department of Food andutrition at Purdue University, Westafayette, IN. “They all have mea-urement errors! It is just in differentegrees.”
RIGINShe beginnings of the FFQ can beraced to Bertha Burke, PhD. In 1947he Journal published her diet his-ory method, a procedure that in-luded a 3-day food intake record, a4-hour recall, and an accounting ofhe frequency of food intakes over aeriod of 1 to 3 months (4,5).The multilayered method was later
eparated into individual tools (6),hough Kristal and colleagues credit a981 cancer study by Sir Richardoll, PhD and Sir Richard Peto, PhD,ith the official birth of the FFQ. “Al-ost all early case-control studies
sed a technique called ‘diet history’o assess either current or past diet.iet histories were lengthy, open-nded, and unstandardized inter-iews administered and analyzed byutritionists, which attempted toharacterize what a study participantusually’ ate,” Kristal and colleaguesaid. “When used in case-control stud-es, diet histories are highly subject toias, given the difficulty and rarity ofnterviewer (and, of course, partici-ant) blinding. And because the inter-iew could take 90 minutes, it wasnsuitable for large cohort studies.hus, out of necessity was born a sim-lified, self-administered, and inex-ensive form of the diet history, theood frequency questionnaire” (2).
“FFQs as we currently know themeally started in earnest in the 1980sith the Harvard and Block instru-ents that were widely available and
ad analytic software associated withhem,” says Amy F. Subar, PhD,PH, RD, nutritionist for the Na-
ional Cancer Institute, Bethesda,
D. “There were earlier ones, but tion Journal
hese comprehensive FFQs wereeadily available and easily ana-yzed.”
IMITATIONS OF THE FFQhe low cost and ease-of-use of theverage FFQ also brings about theiggest criticism of the tool: it is tooague. “Some questionnaires aren’too precise. Did you drink a smalluantity of fizzy drink, a mediumuantity, or a large one? My small isalf-a-glass. How small is yourmall?” said Stephen Daniells, PhD, aood science reporter in a Food Navi-ator editorial. “It would seem to beo entirely obvious that such resultsre less than accurate that manyould wonder why scientists are col-
ecting them at all” (3).Subar, who helped create the Dietistory Questionnaire at the Na-
ional Cancer Institute (7), says thatFQs do have their share of prob-
ems. “FFQs are prone to measure-ent error. Cognitively, the usual fre-
uency of intake questions is difficulto answer. The number of foods onean ask about is limited and extensiveetail about food preparation is notollected. FFQs generally query usualortion size, which may not be soroblematic for discreet foods likeieces of fruit, but can be quite diffi-ult and highly variable for foods likeasta, vegetables, beverages, andeats.”Subar says that things get more
omplex in varied, multi-ingredientoods like casseroles or lasagna. FFQood types are based on the generalopulation. “For example, the nutri-nt values assigned for lasagna wille the same for those who make itrom scratch and for those who buy itre-prepared,” Subar says. “However,FQs are often the only practicalethod available to researchers do-
ng large-scale nutrition research ande do the best we can to design them
o capture usual dietary intakes.”
of the AMERICAN DIETETIC ASSOCIATION 1541
ioswte“oa
rtgbs[ew
RFwliragFl
coscfffiods
aFFHcstdmwf2cngec
wwcPtstrt
R1
2
3
4
5
6
7
BEYOND THE HEADLINES
1
A U T H O R G U I D E L I N E S
If you would like to
submit an article to
the Journal, author
guidelines can be
found in the January
2006 issue
(J Am Diet Assoc.
2006;105:140-147)
or on the Journal
submission Web site
at www.editorial
manager.com/adaj.
l
542 October 2006 Volume 106 Number 10
The other challenge is underreport-ng, following the maxim that some-ne behaves differently when ob-erved. “We have people quantifyinghat they eat and just having to do
hat will cause people to change theirating habits,” Champagne says.We’re not trying to make a judgmentf what they eat, but of the other vari-bles we may be looking at.”For instance, Champagne says, if
esearchers are examining alcohol in-ake, “someone might have fivelasses of wine or a couple six packs,ut he or she wouldn’t want that tohow up on the food intake record.Unfortunately], if we don’t get hon-st info, we have missing data—ande don’t know we’re missing data.”
ELIABLE SOLUTIONSFQ-based research can be balancedith other tools. For instance, doubly
abeled water (water that contains ansotope for tracking purposes) enablesesearchers to measure average met-bolic rate. Researchers can seereater questionnaire accuracy sinceFQs measure intake while doubly
abeled water measures output.“Many of our conclusions regarding
ancer or heart disease are the resultsf using FFQs in large populations,”ays Champagne. “However, when aonsensus is reached, it’s not justrom the FFQ but also data we haverom animal models and data we haverom well-controlled clinical stud-es. . . FFQs are used in addition tother parts of science. No one studyesign is going to give us all the an-wers.”While critical of the tool, Kristal
nd colleagues acknowledged that theFQ was financially appealing. TheFQ was used for the Women’sealth Initiative despite its statisti-
al limitations due to economic con-traints. The Women’s Health Initia-ive considered FFQ alternatives andetermined that the dietary assess-ent at baseline (for 160,000 women)ould cost $23.2 million for 3-day
ood records and $25 million for three4-hour recalls. Using FFQs wouldost $1.2 million (2). In hindsight,ow with 15 years completed and lin-ering questions regarding adher-nce, the cost/benefit might have beenonsidered in greater detail.“To redesign the FFQ to get the
evel of detail offered by another tool
ould make it a totally different toolith different research burden and
ause,” says Maureen A. Murtaugh,hD, RD, associate professor of Nu-rition Epidemiology at the Univer-ity of Utah, Salt Lake City. “I thinkhe issue is choosing it in the rightesearch settings and interpretinghe information accordingly.”
eferences. Subar AF, Kipnis V, Troiano RP,
Midthune D, Schoeller DA, Bing-ham S, Sharbaugh CO, Trabulsi J,Runswick S, Ballard-Barbash R,Sunshine J, Schatzkin A. Using in-take biomarkers to evaluate theextent of dietary misreporting in alarge sample of adults: The OPENStudy. Am J Epidemiol. 2003;158:1-13.
. Kristal AR, Peters U, Potter JD. Isit time to abandon the food fre-quency questionnaire? Cancer Epi-demiol Biomarkers Prev. 2005;14:2826-2828.
. Daniells S. Time to ditch theFFQ. Food Navigator. January30, 2006. Available at: http://www.foodnavigator.com/news/ng.asp?id�65448. Accessed May 25,2006.
. Burke BS. The dietary history as atool in research. J Am Diet Assoc.1947;23:1041-1046.
. Committee on Diet and Health,Food and Nutrition Board, Commis-sion on Life Sciences, National Re-search Council. Diet and Health:Implications for Reducing ChronicDisease Risk. Washington, DC: Na-tional Academy Press; 1989:24.
. Heady JA. Diets of bank clerks:Development of a method of classi-fying the diets of individuals foruse in epidemiologic studies. J RStat Soc. 1961;124:336-371.
. Thompson FE, Subar AF, BrownCC, Smith AF, Sharbaugh CO,Jobe JB, Mittl B, Gibson JT,Ziegler RG. Cognitive research en-hances accuracy of food frequencyquestionnaire reports: results ofan experimental validation study.J Am Diet Assoc. 2002;102:212-225.