reproducibility of reported nutrient intake and supplement use during a past pregnancy: a report...

Reproducibility of reported nutrient intake and supplement useduring a past pregnancy: a report from the Children’s OncologyGroup

Jaclyn L. F. Boscoa, Marilyn Tsengb, Logan G. Spectorc, Andrew F. Olshand, and Greta R.Bunine,fa Department of Epidemiology, Boston University School of Public Health, 715 Albany Street, T3East, Boston, MA 02118b Population Science Division, Fox Chase Cancer Center, 333 Cottman Avenue, Philadelphia, PA19111c Division of Epidemiology and Clinical Research, Department of Pediatrics, University ofMinnesota, 420 Delaware Street, SE, Minneapolis, MN 55455d Department of Epidemiology, School of Public Health, University of North Carolina, 2101BMcGavran-Greenberg Hall, 135 Dauer Drive, Chapel Hill, NC 27599e Division of Oncology, The Children’s Hospital of Philadelphia, 3535 Market Street, Philadelphia,PA 19104f Department of Pediatrics, University of Pennsylvania School of Medicine, Philadelphia, PA19104

SummaryMaternal nutrition and diet have been suggested to play a role in many conditions of her offspring.Studies using food frequency questionnaires (FFQs) have reported associations with maternal diet,but these findings are difficult to interpret because the reliability and validity of the FFQs for dietduring a past pregnancy are not known. We determined the reproducibility of reported diet andsupplement use during a past pregnancy in a subset of mothers interviewed for a case-controlstudy of maternal diet in relation to risk of childhood brain tumors. Cases were Children’sOncology Group patients, diagnosed at age <6 with medulloblastoma or primitiveneuroectodermal tumor from 1991–1997. Area code, race/ethnicity, and birth date matchedcontrols were selected by random-digit-dialing. Case and control mothers completed a modifiedWillett FFQ a mean of five years after the index child’s birth. A mean of 3.6 months later, a subsetof mothers consisting of 52 case and 51 control mothers repeated the interview and comprise thereproducibility study population. The mean intra class correlation (ICC) was 0.59 (range 0.41,0.69) for energy- adjusted nutrients from dietary sources only; it was 0.41 (range −0.06, 0.70)when supplements were included. Agreement for reporting multivitamin use during pregnancy bytime period and pattern was good to very good (kappa= 0.66 to 0.85). Overall, the reproducibilityof nutrient estimates and supplement use in pregnancy was good and similar to that reported foradult diet.

Corresponding author: Dr. Greta R Bunin, Children’s Hospital of Philadelphia, 3535 Market St, Rm 1472, Philadelphia, PA 19104;Phone: 215 590-1445, Fax 215 590-3020, [email protected].

NIH Public AccessAuthor ManuscriptPaediatr Perinat Epidemiol. Author manuscript; available in PMC 2011 March 8.

Published in final edited form as:Paediatr Perinat Epidemiol. 2010 January ; 24(1): 93–101. doi:10.1111/j.1365-3016.2009.01070.x.

NIH

-PA Author Manuscript

NIH


NIH


IntroductionMaternal nutrition and diet have been suggested to play a role in many common conditionsof her offspring including allergy, asthma, cognitive development, and obesity, 1–4 withmost findings coming from cohort studies. Maternal nutrition may also affect the risk of lesscommon diseases of childhood, thus retrospective assessment of diet using case-controldesigns will be necessary. For diseases such as childhood cancers, because they are rare andoccur years after birth, researchers must usually use a case-control study design and rely onmaternal recall. In case-control studies, aspects of mothers’ diet and supplement use duringpregnancy have been associated with the child’s risk of cancer, specifically multivitaminswith lower risk of brain tumors,5–8 leukemia,9 and neuroblastoma;10 infrequent eating ofvegetables with higher risk of retinoblastoma;11 frequent eating of foods containing DNAtoposiomerase II inhibitors with higher risk of specific subgroups of infant leukemia;12, 13

and frequent eating of cured meats with higher risk of brain tumors.14–17 However, theveracity of the findings depends on the reproducibility and validity of the retrospectivelycollected dietary data.

The dietary information (except for supplement use) in these studies was collected usingfood frequency questionnaires (FFQs),18 the primary method for retrospectively assessingdiet in case-control studies. FFQs have been found to be reasonably reproducible and validto assess past adult diet,19–22 but their limitations are also well-known,23 and their accuracyfor assessing diet during a past pregnancy is not well studied. Ideally, pregnant womenwould complete a non-FFQ dietary assessment, such as several 24 hour recalls, and thencomplete the FFQ several years later. Food and nutrient intake would be compared using thedietary assessment that took place during the pregnancy as the ‘gold standard.’ However, thesituation of a large enough group of women whose diet was studied during pregnancy andwho are available for study several years later is rare. In the absence of this ideal situation,other comparisons can be made that begin to address the accuracy of reporting diet during apast pregnancy. However, we know of only our previous study and one other that evaluatedreproducibility of FFQs for recalled pregnancy diet.24, 25 We know of no previous studieson the reproducibility of reported supplement use during pregnancy and no studies that havedirectly assessed validity of reported diet or supplement use during a past pregnancy.

In our previous study of FFQ reproducibility, we concluded that pregnancy diet was recalledabout as well as adult diet generally.24 In this study, women who had completed FFQsduring pregnancy were re-contacted 3–7 years later and asked to complete the same self-administered FFQ or a different FFQ administered as a telephone interview. However, thesample represented a rather select group, limiting generalizability of these results to a morediverse population: Two-thirds of the women were college graduates, 97% were white, andall had completed multiple FFQs during their pregnancies as part of their participation in aresearch study. Further, our previous analysis did not report on the reproducibility ofreported supplement use. To extend the findings of our previous analysis, we performed asecond study of reproducibility, in which mothers who had completed a FFQ as part of theirparticipation in a national case-control study of childhood brain tumors were re-interviewedthree months later. Here, we report on the reproducibility of their reported nutrient intakeand supplement use during pregnancy.

MethodsStudy Population

A case-control study of medulloblastoma (MB) and primitive neuroectodermal tumor(PNET) was conducted to investigate the role of maternal diet during pregnancy.26 Eligiblepatients were diagnosed with MB/PNET in the brain before age six, between 1991 and 1997,

Bosco et al.

Paediatr Perinat Epidemiol. Author manuscript; available in PMC 2011 March 8.

NIH


NIH


NIH


and registered with the Children’s Oncology Group (COG) through institutions previouslyaffiliated with the Children’s Cancer Group (CCG). (Text deleted)

Controls were identified through random-digit-dialing (RDD) and individually matched tocases on area code, child’s race/ethnicity (non-Hispanic white, non-Hispanic black, other)and child’s date of birth (within six months for cases with age of diagnosis <1 year, within 1year for all others). (Text deleted) Details about the main study’s exclusions have beenreported previously. 26 The main study population consisted of 315 case-control pairs.

Participants in the reproducibility study were drawn from the 146 mothers of cases andcontrols who completed the telephone interview for the main study between September 1996and June 1997. Three months after completing the interview, each of these mothers was senta letter asking her to repeat the interview and offered $20. Of the 146 mothers eligible forthe sub-study, 104 (71%) completed the second interview. The reasons for non-participationwere refusal, inability to locate and contact by phone, and inability to interview before studyended. One case mother was excluded from the analysis because of missing data. The finalsample consisted of 52 case mothers and 51 control mothers, including 17 matched case-control pairs.

We obtained approval from the institutional review boards of all participating institutionsand all participants provided informed consent.

Data CollectionTrained personnel conducted telephone interviews that included a modified Willett FFQ18

with 112 items to assess diet during two time periods: the year prior to pregnancy and duringthe second trimester of pregnancy. This analysis examined nutrient data from the secondtrimester.

Mothers were also asked about their use of supplements (multivitamins, vitamin A, vitaminC, vitamin E, calcium, selenium, iron, zinc, and folate) in the year before and duringpregnancy. Supplement users were asked about duration and frequency of use, if they weretaking the supplement when they became pregnant, brand and type (multivitamins), dose(individual supplements), month during pregnancy when use began and ended, prescriptionor non-prescription (multivitamins). If the mother remembered only part of the brand/type ofprenatal multivitamin, the interviewer read common product names containing theremembered words from a list to prompt further recall.

The micronutrient content of multivitamins was assigned based on brand/type.18 When thebrand/type was not known or not in the database, the multivitamin was categorized asprescription or non-prescription, prenatal or regular, based on the answers to the relevantquestions. A prescription prenatal multivitamin of unknown brand and a non-prescriptionprenatal multivitamin of unknown brand were assigned micronutrient amounts typical in theyears of the index pregnancies.27–29 For single micronutrient supplements, we used the dosereported by the mother or, if not known, a typical dose (500 mg vitamin C, 400 InternationalUnits (IU) vitamin E, 400 μg folate, 60 mg iron, and 200 mg calcium).

To determine whether a supplement was used close to conception, before most women knewthey were pregnant, we asked mothers who reported use in the year before the pregnancy ifthey were taking the supplement when they became pregnant. Mothers who answered yeswere considered to have used the supplement close to conception. Some mothers who didnot report use in the year before pregnancy stated they were taking the supplement beforethe pregnancy began, in response to the question about when during the pregnancy usestarted. These mothers were considered to have close to conception usage.

Bosco et al.


NIH


NIH


NIH


Statistical AnalysesAnalyses were performed for case and control mothers combined and separately using SASversion 9.1 software (SAS Institute Inc., Cary, North Carolina) and SPSS Base 12.0 (SPSSInc., Chicago, Illinois). Daily intake of selected micronutrients from food was calculated bysumming the content across all foods. The Willett FFQ standard portion18 was assumed asportion size was not collected. Nutrient intake was estimated from food only and from foodplus supplements and the estimates were energy-adjusted using the residual method.30

Reproducibility of nutrient estimates—Intra-class correlations (ICC) and 95%confidence intervals (CI) were calculated to assess the reproducibility of estimates ofnutrient intake from the FFQs in the first (FFQ1) and second (FFQ2) interviews. Thesemeasures were calculated for crude and energy-adjusted estimates of intake from food onlyand food plus supplements.

We also assessed the reproducibility of quartiles of nutrient intake using the weighted kappacoefficient, which accounts for the ordinality of the quartiles. Quartile was assigned basedon the distribution of nutrient intake among case and control mothers combined for analysesof all mothers and based on the distribution of case mothers only and control mothers onlyfor separate analyses. Exact and adjacent quartile agreement was calculated for each nutrient(crude and energy-adjusted) using FFQ1 as the reference. The observed quartile agreementwas compared to the expected agreement of 0.25 for exact quartile agreement and 0.625 foradjacent quartile agreement using a one-sample proportion test with alpha set at 0.001because of the large number of statistical tests performed.

Reproducibility of reported supplement use—The agreement of reportedsupplement use between the two questionnaires was assessed by kappa coefficients.Multivitamin use at any time during the pregnancy, use during each portion of pregnancy(around conception, 1st trimester, 2nd trimester, and 3rd trimester), and patterns of use (nevertook multivitamins, started taking multivitamins around conception and continued untildelivery, started taking during the 1st trimester and continued until delivery, and startedtaking during the 2nd trimester and continued until delivery, other) were analyzed. For singlenutrient supplements, only any use during the pregnancy was analyzed because of the smallnumber of users.

ResultsTable 1 shows descriptive characteristics of the reproducibility study population (n=103).Case and control mothers differed only in ethnicity (Hispanic: 25% of cases vs. 4% ofcontrols, p=0.02). As the results in this report did not change appreciably when the analyseswere limited to non-Hispanic whites, we present the results for all races/ethnicitiescombined. Additionally, no systematic differences between case and control mothers wereobserved and, thus, only results for the combined group of 103 mothers are presented. Theoriginal interview took place, on average, 5.0 years (range 0.3, 9.4; SD=2.1) after the indexchild’s birth and the second interview 3.6 months (range 2.3, 8.6; SD=1.1) later.

The reproducibility study population of 52 case mothers and 51 control mothers was similarto the main study population of 315 cases and 315 controls in child’s age at first interview,race and education except that proportionately, there were more Hispanic case mothers inthe subgroup (25% vs. 10%).

Bosco et al.


NIH


NIH


NIH


Nutrients without supplementsThe mean ICC for crude nutrients without supplements was 0.62 (range 0.45, 0.72) (Table2). Overall, the ICCs for energy-adjusted nutrient estimates were slightly lower with a meanof 0.59 (range 0.41, 0.69). (Text deleted)

The exact quartile agreement of energy-adjusted nutrients without supplements ranged from39% to 54% and kappa coefficients ranged from 0.32 to 0.57 (Table 3). The agreements forall nutrients except polyunsaturated fat and vitamin B6 were statistically significantlydifferent from the expected of 0.25 at α=0.001. Seventy-seven to 91% of mothers fell intothe same quartile or the adjacent quartile for both FFQs. The adjacent agreements for allnutrients except polyunsaturated fat and linoleic acid were statistically significantly differentfrom the expected of 0.625 at α=0.001. Results were similar for crude nutrients (not shown).(Text deleted)

Nutrients with supplementsThe mean ICC for crude and energy-adjusted nutrients with supplements was 0.47 (range−0.03, 0.69) and 0.41 (range −0.06, 0.70), respectively (Table 2). With supplements,energy-adjusted vitamin E (ICC= −0.02, 95% CI: −0.21, 0.17) and iron (ICC= −0.06, 95%CI: −0.25, 0.13) had lower ICC than without supplements and the lowest ICC of allnutrients. Without vitamin E and iron, the mean ICC for nutrients with supplements was0.49 (range 0.33, 0.70), somewhat lower than for nutrients without supplements. (Textdeleted)

We performed analyses of subsets of mothers to determine whether inconsistent reporting ofsupplement use contributed to the low reliability for vitamin E and iron. For vitamin E,intake including supplements was reproducible for mothers who reported the samemultivitamin in both interviews (n=62, ICC=0.71) but not for those who reported differentmultivitamins (n=39, ICC= −0.08). Four mothers who reported vitamin E supplement usewere excluded from the subset analyses because of their small number. Similar analyses foriron intake with supplements did not reveal an interpretable pattern.

Exact quartile agreement for energy-adjusted nutrients with supplements ranged from 31%to 59% with kappa coefficients ranging from 0.06 to 0.59 (Table 3), and exact agreementsfor all except vitamin E and calcium were statistically significantly different from theexpected of 0.25 at α=0.001. The range for adjacent agreement was 63% to 91%, with allstatistically significantly different from the expected of 0.625 (α=0.001) except vitamin E,calcium, and zinc. (Text deleted) Kappa coefficients were similar for crude (not shown) andenergy-adjusted nutrients (Table 3).

Multivitamin useAgreement for reporting any multivitamin use was 0.71 (Table 4). Agreement ranged from0.66 to 0.85 for time period of use and from 0.73 to 0.75 for the two common patterns ofuse. Of multivitamin users, 63% (60 of 95) consistently reported the same brand/type. (Textdeleted) Individual micronutrient supplements

Agreement for use of calcium and iron supplements was 0.61 and 0.65, respectively (Table4). No mothers reported taking beta-carotene, vitamin A, or selenium and too few reportedtaking vitamin C, vitamin E, or folic acid to permit separate analyses.

Most mothers who took individual supplements did not remember their dose. For example,six of the 27 iron supplement users reported a dose in FFQ1 and four of 22 in FFQ2. Onlyone mother reported the same dose in both questionnaires.

Bosco et al.


NIH


NIH


NIH


DiscussionThis study is one of only a few to evaluate the reproducibility of self-reported dietary intakeduring pregnancy,24, 25 and the first to our knowledge to evaluate the reproducibility of self-reported supplement intake during pregnancy. We found that ICCs for energy-adjustednutrient intake from food alone averaged 0.59 (range 0.41, 0.69), while the average ICC fornutrients including supplements was lower, 0.41, with a wider range (−0.06, 0.70) becauseof low correlations for vitamin E and iron. We also found good reproducibility for reportingof use of multivitamins and of iron and calcium supplements, although estimatingsupplement dosages was difficult.

Our ICCs for nutrients without supplements are similar to those from the one study ofrecalled pregnancy diet (range 0.4, 0.8)25 and to those from studies of the general adultpopulation (range 0.5, 0.7).22 Our results for nutrients with supplements are also comparableto those in the general adult population except for vitamin E and iron, particularly afterenergy adjustment. The low correlations for vitamin E with supplements occurred amongmothers who reported different multivitamins in the two interviews. The vitamin E contentof prenatal multivitamins varied from 0 to 30 I.U. and was generally high compared to themedian intake from food (9 I.U.) in this population. Therefore, multivitamins made a largecontribution to vitamin E content and the inconsistent reporting of brand/type may explainthe poor reliability. For iron, the explanation for the poor reliability is less certain. It may berelevant that over 20% of mothers reported taking a separate iron supplement and that thevariability in iron content of multivitamins was large (40 mg to 120 mg) compared to themedian intake from food (14 mg). The fact that the crude ICC for iron with supplements wassimilar to those for other nutrients and dropped only with energy adjustment suggestsmisreporting of iron supplement intake in relation to energy intake, an observation wecannot explain. The moderately high quartile agreement for energy-adjusted iron withsupplements suggests that the low ICC may be due to misreporting by a small proportion ofmothers.

The reproducibility of reported multivitamin use was high for any use (k=0.71); for mosttime periods and common patterns of use, the agreement was good to very good (range 0.66,0.85).31, 32 We know of no data on a past pregnancy with which to compare our results, butsome data are available on non-pregnant adults. In a study of older adults, participantscompleted a questionnaire twice, three months apart, concerning their supplement use in thelast 10 years;33 the observed agreement was good (k=0.78) for average frequency ofmultivitamin use. Although not about pregnancy, the interval of recall and the finding thatmultivitamin use is reproducibly reported were similar to ours.

Iron and calcium had good reliability for any use of these supplements (k=0.61 and k=0.65,respectively). In another study of older adults, Murphy et al. observed similar levels ofreproducibility for commonly used single nutrient supplements,34 although the participantsreported their use over the past year, unlike our study.

The mean interval between FFQs was 3.6 months, thus, our results likely reflect thereproducibility of the mothers’ reported diet and not simply their ability to recall theirprevious responses.22 The reproducibility of estimates of intake including supplements,however, may be overestimated for some micronutrients. Mothers who did not rememberthe brand/type of multivitamin they took and mothers who did not remember the dose ofindividual supplements they took were assigned doses typical of the reported products. Forexample, a mother who did not remember the brand and name of the multivitamin she tookwas assigned the same typical doses for both questionnaires. By assigning typical doses

Bosco et al.


NIH


NIH


NIH


when the dose was unknown, we have overestimated the similarity of the estimates from thetwo questionnaires.

Our findings indicate that estimating dosage for supplements in a past pregnancy isproblematic. In childhood cancer case-control studies, mothers are asked to report theirsupplement use during a past pregnancy, at a time when most are not pregnant, unlikely tobe using the same product as during pregnancy, and therefore unable to obtain the brand/type and dose from the label. This is clearly a disadvantage in obtaining accurate dataretrospectively. In our sample, less than one-third of mothers reporting use of singlesupplements could provide a dose and even fewer reported the same dose in bothquestionnaires. Thus, our findings suggest that questionnaires on supplement use during apregnancy several years in the past need not ask about dose because it cannot be reportedreliably; assigning a standard dose would seem the better approach. Other potential sourcesof error in reporting supplement use deserve discussion. Some mothers may haveinconsistently reported their use of individual supplements because they did not understandthat we were asking about supplements apart from multivitamins. Another concern isdetermining the brand and type of multivitamin used. The quantities of nutrients can varywidely between the many brands and individual products within brands of prenatalmultivitamins, both over-the-counter and prescription. As a result, misreporting a brand orwithin-brand type can lead to substantial error in estimating dose. In our study, for example,misreporting a brand or type greatly reduced correlations for vitamin E. Providingphotographs of the bottles and the pills of the most commonly used products might improvereporting.

Differences between case and control mothers in reliability of reporting could introducedifferential bias into studies. However, our sample size was not large enough to assess theimportant question of case-control differences. We hope that future studies with largersample sizes address this issue.

Neither our study nor any previous studies have directly assessed validity of recall of dietduring a past pregnancy. Such a study would require collection of dietary information usinga referent method during the pregnancy. In our previous work, we compared women’s recallof their diet during a pregnancy 3–7 years in the past with their diet as they reported itduring the pregnancy.24 We found that women recalled their diet during pregnancy withsimilar accuracy as in the general adult population.24 Other studies have reported reasonablevalidity of FFQ data on diet as long as 10–15 years in the past.19–22 Overall, these studiesprovide indirect support for the validity of FFQ data in studies on maternal diet duringpregnancy.

As more epidemiologic studies of childhood cancer and other conditions focus on diet andsupplement use during pregnancy, the need for data directly evaluating the performance ofthe dietary questionnaires is growing. Our data reported here contribute to the sparseliterature on the performance of a FFQ to assess diet during a past pregnancy. We observedlevels of reproducibility for most nutrients similar to those seen for adult diet generally.Difficulty in reporting brand and dosage information for multivitamins may havecontributed to reduced correlations for vitamin E, but did not explain the low ironcorrelations. Mothers’ use of multivitamins, including timing and patterns of use, andcommon individual supplements was reliably reported. Our results indicate that data onmaternal intake of most nutrients and common supplements during a past pregnancy can bereproducibly collected in case-control studies of childhood cancers and other rare childhooddiseases.

Bosco et al.


NIH


NIH


NIH


AcknowledgmentsThis work was supported by grant CA60951 from the National Cancer Institute and by grant P30ES10126 from theNational Institute of Environmental Health Sciences. Funding support was also provided by the COG Grant U10CA 98543. A complete listing of grant support for research conducted by CCG and POG before initiation of theCOG grant in 2003 is available online at: http://www.childrensoncologygroup.org/admin/grantinfo.htm The authorsthank the project managers and interviewers (Kathy Walsh, Anne Goldblatt, the late Jean Rodwell, Mary Rewinski,Christine Plourde, and Sallie McLaughlin) for their hard work and dedication during the data collection phase of thestudy.

References1. Litonjua AA, Rifas-Shiman SL, Ly NP, Tantisira KG, Rich-Edwards JW, Camargo CA Jr, et al.

Maternal antioxidant intake in pregnancy and wheezing illnesses in children at 2 y of age. AmericanJournal of Clinical Nutrition 2006 Oct;84(4):903–911. [PubMed: 17023719]

2. Daniels JL, Longnecker MP, Rowland AS, Golding J. Fish intake during pregnancy and earlycognitive development of offspring. Epidemiology 2004 Jul;15(4):394–402. [PubMed: 15232398]

3. Yajnik C. Nutritional control of fetal growth. Nutrition Reviews 2006 May;64(5 Pt 2):S50–51.discussion S72–91. [PubMed: 16770953]

4. Krechowec SO, Vickers M, Gertler A, Breier BH. Prenatal influences on leptin sensitivity andsusceptibility to diet-induced obesity. Journal of Endocrinology 2006 May;189(2):355–363.[PubMed: 16648302]

5. Bunin GR, Kuijten RR, Buckley JD, Rorke LB, Meadows AT. Relation between maternal diet andsubsequent primitive neuroectodermal brain tumors in young children. New England Journal ofMedicine 1993;329:536–541. [PubMed: 8336753]

6. Preston-Martin S, Pogoda JM, Mueller BA, Lubin F, Holly EA, Filippini G, et al. Prenatal vitaminsupplementation and risk of childhood brain tumors. International Journal of Cancer Suppl1998;11:17–22.

7. Preston-Martin S, Pogoda JM, Mueller BA, Lubin F, Modan B, Holly EA, et al. Results from aninternational case-control study of childhood brain tumors: the role of prenatal vitaminsupplementation. Environmental Health Perspectives 1998;106(S3):887–892. [PubMed: 9646053]

8. Bunin G, Gallagher PR, Rorke-Adams LB, Robison LL, Cnaan A. Maternal supplement,micronutrient, and cured meat intake during pregnancy and risk of medulloblastoma duringchildhood: a Children’s Oncology Group study. Cancer Epidemiology Biomarkers & Prevention2006;15(9):1660–1667.

9. Ross JA, Blair CK, Olshan AF, Robison LL, Smith FO, Heerema NA, et al. Periconceptionalvitamin use and leukemia risk in children with Down syndrome: a Children’s Oncology Groupstudy. Cancer 2005 Jul 15;104(2):405–410. [PubMed: 15952191]

10. Olshan AF, Smith JC, Bondy ML, Neglia JP, Pollock BH. Maternal vitamin use and reduced riskof neuroblastoma. Epidemiology 2002;13:575–580. [PubMed: 12192228]

11. Orjuela MA, Titievsky L, Liu X, Ramirez-Ortiz M, Ponce-Castaneda V, Lecona E, et al. Fruit andvegetable intake during pregnancy and risk for development of sporadic retinoblastoma. CancerEpidemiology Biomarkers & Prevention 2005 Jun;14(6):1433–1440.

12. Spector LG, Xie Y, Robison LL, Heerema NA, Hilden JM, Lange B, et al. Maternal diet and infantleukemia: the DNA topoisomerase II inhibitor hypothesis: a report from the children’s oncologygroup. Cancer Epidemiology Biomarkers & Prevention 2005 Mar;14(3):651–655.

13. Ross JA, Potter JD, Reaman GH, Pendergrass TW, Robison LL. Maternal exposure to potentialinhibitors of DNA topoisomerase II and infant leukemia (United States): a report from theChildren’s Cancer Group. Cancer Causes & Control 1996 Nov;7(6):581–590. [PubMed: 8932918]

14. Preston-Martin S, Yu MC, Benton B, Henderson BE. N-Nitroso compounds and childhood braintumors: a case-control study. Cancer Research 1982;42:5240–5245. [PubMed: 7139628]

15. Kuijten RR, Bunin GR, Nass CC, Meadows AT. Gestational and familial risk factors for childhoodastrocytoma: results of a case-control study. Cancer Research 1990;50:2608–2612. [PubMed:2328486]

Bosco et al.


NIH


NIH


NIH


http://www.childrensoncologygroup.org/admin/grantinfo.htm

16. Preston-Martin S, Pogoda JM, Mueller BA, Holly EA, Lijinsky W, Davis RL. Maternalconsumption of cured meats and vitamins in relation to pediatric brain tumors. CancerEpidemiology Biomarkers & Prevention 1996;5:599–605.

17. Baldwin RT, Preston-Martin S. Epidemiology of brain tumors in childhood--a review. Toxicologyand Applied Pharmacology 2004 Sep 1;199(2):118–131. [PubMed: 15313584]

18. Willett, W. Food frequency methods. In: Willett, W., editor. Nutritional Epidemiology. 2. NewYork: Oxford University Press; 1998. p. 74-100.

19. van Leeuwen FE, de Vet HCW, Hayes RB, van Staveren WA, West CE, Hautvast JGAJ. Anassessment of the relative validity of retrospective interviewing for measuring dietary intake.American Journal of Epidemiology 1983;1983:752–758. [PubMed: 6638002]

20. Willett WC, Sampson L, Browne ML, Stampfer MJ, Rosner B, Hennekens CH, et al. The use of aself-administered questionnaire to assess diet four years in the past. American Journal ofEpidemiology 1988;127:188–199. [PubMed: 3337073]

21. Sobell J, Block G, Koslowe P, Tobin J, Andres R. Validation of a retrospective questionnaireassessing diet 10–15 years ago. American Journal of Epidemiology 1989 Jul;130(1):173–187.[PubMed: 2741904]

22. Willett, WC.; Lenart, E. Reproducibility and validity of food-frequency questionnaires. In: Willett,W., editor. Nutritional Epidemiology. Vol. Chapter 6. New York, New York: Oxford UP; 1998. p.101-47.

23. Byers T. Food frequency dietary assessment: how bad is good enough? American Journal ofEpidemiology 2001 Dec 15;154(12):1087–1088. [PubMed: 11744510]

24. Bunin GR, Gyllstrom ME, Brown JE, Kahn EB, Kushi LH. Recall of diet during a past pregnancy.American Journal of Epidemiology 2001;154:1136–1142. [PubMed: 11744519]

25. Wilkins JR, Bunn JY. Comparing dietary recall data for mothers and children obtained on twooccasions in a case-control study of environmental factors and childhood brain tumors.International Journal of Epidemiology 1997;26:953–963. [PubMed: 9363515]

26. Bunin GR, Kushi LH, Gallagher PR, Rorke-Adams LB, McBride ML, Cnaan A. Maternal dietduring pregnancy and its association with medulloblastoma in children: a Children’s OncologyGroup study. Cancer Causes & Control 2005;16:877–891. [PubMed: 16132798]

27. Physician’s Desk Reference. 39. Oradell, NJ: Medical Economics Co; 1985.28. Physicians’ Desk Reference for Nonprescription drugs. 16. Oradell, NJ: Medical Economics Co;

1995.29. Drug Facts and Comparisons. St. Louis: Facts and Comparisons, Wolters Kluwer; 1996.30. Kushi LH, Sellers TA, Potter JD, Nelson CL, Munger RG, Kaye SA, et al. Dietary fat and

postmenopausal breast cancer. Journal of the National Cancer Institute 1992;84:1092–1099.[PubMed: 1619683]

31. Byrt T. How good is that agreement? Epidemiology 1996 Sep;7(5):561. [PubMed: 8862998]32. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics

1977 Mar;33(1):159–174. [PubMed: 843571]33. Satia-Abouta J, Patterson RE, King IB, Stratton KL, Shattuck AL, Kristal AR, et al. Reliability and

validity of self-report of vitamin and mineral supplement use in the vitamins and lifestyle study.American Journal of Epidemiology 2003 May 15;157(10):944–954. [PubMed: 12746248]

34. Murphy SP, Wilkens LR, Hankin JH, Foote JA, Monroe KR, Henderson BE, et al. Comparison oftwo instruments for quantifying intake of vitamin and mineral supplements: a brief questionnaireversus three 24-hour recalls. American Journal of Epidemiology 2002 Oct 1;156(7):669–675.[PubMed: 12244036]

Bosco et al.


NIH


NIH


NIH


https://www.researchgate.net/publication/40164534_An_assessment_of_the_relative_validity_of_retrospective_interviewing_for_measuring_dietary_intake?el=1_x_8&enrichId=rgreq-53d2d079afd3b57572f0d88eba227e0b-XXX&enrichSource=Y292ZXJQYWdlOzQxMDM4NTAyO0FTOjEwMTg4ODMwOTU5NjE2N0AxNDAxMzAzNDc0OTMw



https://www.researchgate.net/publication/11611429_Food_Frequency_Dietary_Assessment_How_Bad_Is_Good_Enough?el=1_x_8&enrichId=rgreq-53d2d079afd3b57572f0d88eba227e0b-XXX&enrichSource=Y292ZXJQYWdlOzQxMDM4NTAyO0FTOjEwMTg4ODMwOTU5NjE2N0AxNDAxMzAzNDc0OTMw

https://www.researchgate.net/publication/11611429_Food_Frequency_Dietary_Assessment_How_Bad_Is_Good_Enough?el=1_x_8&enrichId=rgreq-53d2d079afd3b57572f0d88eba227e0b-XXX&enrichSource=Y292ZXJQYWdlOzQxMDM4NTAyO0FTOjEwMTg4ODMwOTU5NjE2N0AxNDAxMzAzNDc0OTMw

https://www.researchgate.net/publication/21535416_Dietary_Fat_and_Postmenopausal_Breast_Cancer?el=1_x_8&enrichId=rgreq-53d2d079afd3b57572f0d88eba227e0b-XXX&enrichSource=Y292ZXJQYWdlOzQxMDM4NTAyO0FTOjEwMTg4ODMwOTU5NjE2N0AxNDAxMzAzNDc0OTMw



https://www.researchgate.net/publication/14381713_Maternal_consumption_of_cured_meats_and_vitamins_in_relation_to_pediatric_brain_tumors?el=1_x_8&enrichId=rgreq-53d2d079afd3b57572f0d88eba227e0b-XXX&enrichSource=Y292ZXJQYWdlOzQxMDM4NTAyO0FTOjEwMTg4ODMwOTU5NjE2N0AxNDAxMzAzNDc0OTMw



https://www.researchgate.net/publication/20414140_Validation_of_a_Retrospective_Questionnaire_Assessing_Diet_10-15_Years_Ago?el=1_x_8&enrichId=rgreq-53d2d079afd3b57572f0d88eba227e0b-XXX&enrichSource=Y292ZXJQYWdlOzQxMDM4NTAyO0FTOjEwMTg4ODMwOTU5NjE2N0AxNDAxMzAzNDc0OTMw



https://www.researchgate.net/publication/13865379_Comparing_dietary_recall_data_for_mothers_and_children_obtained_on_two_occasions_in_a_case-control_study_of_environmental_factors_and_childhood_brain_tumors?el=1_x_8&enrichId=rgreq-53d2d079afd3b57572f0d88eba227e0b-XXX&enrichSource=Y292ZXJQYWdlOzQxMDM4NTAyO0FTOjEwMTg4ODMwOTU5NjE2N0AxNDAxMzAzNDc0OTMw



https://www.researchgate.net/publication/22310512_The_Measurement_Of_Observer_Agreement_For_Categorical_Data?el=1_x_8&enrichId=rgreq-53d2d079afd3b57572f0d88eba227e0b-XXX&enrichSource=Y292ZXJQYWdlOzQxMDM4NTAyO0FTOjEwMTg4ODMwOTU5NjE2N0AxNDAxMzAzNDc0OTMw

https://www.researchgate.net/publication/22310512_The_Measurement_Of_Observer_Agreement_For_Categorical_Data?el=1_x_8&enrichId=rgreq-53d2d079afd3b57572f0d88eba227e0b-XXX&enrichSource=Y292ZXJQYWdlOzQxMDM4NTAyO0FTOjEwMTg4ODMwOTU5NjE2N0AxNDAxMzAzNDc0OTMw

https://www.researchgate.net/publication/11611437_Recall_of_Diet_during_a_Past_Pregnancy?el=1_x_8&enrichId=rgreq-53d2d079afd3b57572f0d88eba227e0b-XXX&enrichSource=Y292ZXJQYWdlOzQxMDM4NTAyO0FTOjEwMTg4ODMwOTU5NjE2N0AxNDAxMzAzNDc0OTMw

https://www.researchgate.net/publication/11611437_Recall_of_Diet_during_a_Past_Pregnancy?el=1_x_8&enrichId=rgreq-53d2d079afd3b57572f0d88eba227e0b-XXX&enrichSource=Y292ZXJQYWdlOzQxMDM4NTAyO0FTOjEwMTg4ODMwOTU5NjE2N0AxNDAxMzAzNDc0OTMw

https://www.researchgate.net/publication/19819388_The_use_of_a_self-administered_to_assess_diet_four_years_in_the_past?el=1_x_8&enrichId=rgreq-53d2d079afd3b57572f0d88eba227e0b-XXX&enrichSource=Y292ZXJQYWdlOzQxMDM4NTAyO0FTOjEwMTg4ODMwOTU5NjE2N0AxNDAxMzAzNDc0OTMw



https://www.researchgate.net/publication/295314737_Maternal_diet_during_pregnancy_and_its_association_with_medulloblastoma?el=1_x_8&enrichId=rgreq-53d2d079afd3b57572f0d88eba227e0b-XXX&enrichSource=Y292ZXJQYWdlOzQxMDM4NTAyO0FTOjEwMTg4ODMwOTU5NjE2N0AxNDAxMzAzNDc0OTMw



NIH


NIH


NIH


Bosco et al.

Tabl

e 1

Cha

ract

eris

ticM

ean

(SD

)M

in, m

ax%

(n)

Tim

e in

terv

al b

etw

een

FFQ

1 an

d FF

Q2

(mon

ths)

3.6

(1.1

)2.

3, 8

.6

Tim

e in

terv

al b

etw

een

child

’s b

irth

and

FFQ

1 (y

ears

)5.

0 (2

.1)

0.3,

9.4

Mot

her’

s age

at t

ime

of c

hild

’s b

irth

(yea

rs)

29.2

(5.6

)16

.7, 4

0.9

Mot

her’

s rac

e/et

hnic

ity

W

hite

80 (8

3)

H

ispa

nic

14 (1

4)

B

lack

4 (4

)

A

sian

2 (2

)

Mot

her’

s edu

catio

n

Le

ss th

an h

igh

scho

ol12

(12)

H

igh

scho

ol24

(25)

So

me

post

-hig

h sc

hool

29 (3

0)

C

olle

ge g

radu

ate

26 (2

7)

Pr

ofes

sion

al o

r gra

duat

e de

gree

9 (9

)

FFQ

, foo

d fr

eque

ncy

ques

tionn

aire

.


NIH


NIH


NIH


Bosco et al.

Tabl

e 2

Nut

rien

t

With

out s

uppl

emen

tsW

ith su

pple

men

ts

Ene

rgy

adju

sted

ICC

[95%

CI]

Una

djus

ted

ICC

[95%

CI]

Ene

rgy

adju

sted

ICC

[95%

CI]

Una

djus

ted

ICC

[95%

CI]

Cal

orie

s (kc

al)

–0.

63 [0

.50,

0.7

3]

Ani

mal

fat (

g)0.

59 [0

.59,

0.7

0]0.

64 [0

.51,

0.7

4]

Veg

etab

le fa

t (g)

0.63

[0.5

0, 0

.73]

0.70

[0.5

9, 0

.79]

Satu

rate

d fa

t (g)

0.63

[0.4

9, 0

.73]

0.69

[0.5

7, 0

.79]

Mon

osat

urat

ed fa

t (g)

0.67

[0.5

5, 0

.77]

0.68

[0.5

6, 0

.77]

Poly

unsa

tura

ted

fat (

g)0.

42 [0

.24,

0.5

6]0.

64 [0

.51,

0.7

4]

Lino

leic

aci

d (g

)0.

44 [0

.27,

0.5

9]0.

64 [0

.51,

0.7

4]

Ole

ic a

cid

(g)

0.68

[0.5

6, 0

.77]

0.68

[0.5

6, 0

.77]

Om

ega

3-fa

tty a

cids

(g)

0.66

[0.5

4, 0

.76]

0.72

[0.6

1, 0

.80]

Prot

ein

(g)

0.63

[0.4

9, 0

.73]

0.68

[0.4

8, 0

.72]

Ani

mal

pro

tein

(g)

0.59

[0.4

5, 0

.70]

0.61

[0.4

7, 0

.72]

Cho

lest

erol

(mg)

0.65

[0.5

2, 0

.75]

0.62

[0.4

9, 0

.73]

Car

bohy

drat

e (g

)0.

57 [0

.42,

0.6

9]0.

60 [0

.46,

0.7

1]

Die

tary

fibr

e (g

)0.

64 [0

.51,

0.7

4]0.

62 [0

.48,

0.7

2]

Alc

ohol

(g)

0.63

[0.4

9, 0

.73]

0.62

[0.4

9, 0

.73]

Vita

min

A (I

U)

0.51

[0.3

5, 0

.64]

0.58

[0.4

3, 0

.69]

0.52

[0.3

7, 0

.65]

0.58

[0.4

4, 0

.70]

Ret

inol

(IU

)0.

54 [0

.38,

0.6

6]0.

56 [0

.41,

0.6

8]0.

55 [0

.40,

0.6

7]0.

56 [0

.41,

0.6

8]

Thia

min

e (m

g)0.

58 [0

.44,

0.7

0]0.

57 [0

.42,

0.6

8]0.

33 [0

.14,

0.4

9]0.

32 [0

.14,

0.4

9]

Rib

ofla

vin

(mg)

0.54

[0.3

9, 0

.67]

0.63

[0.5

0, 0

.73]

0.42

[0.2

5, 0

.57]

0.46

[0.2

9, 0

.60]

Vita

min

B6

(mg)

0.57

[0.4

2, 0

.69]

0.61

[0.4

8, 0

.72]

0.70

[0.5

8, 0

.78]

0.69

[0.5

8, 0

.78]

Vita

min

B12

(mg)

0.61

[0.4

8, 0

.72]

0.65

[0.5

2, 0

.75]

0.55

[0.4

0, 0

.67]

0.59

[0.4

5, 0

.70]

Vita

min

C (I

U)

0.65

[0.5

3, 0

.75]

0.67

[0.5

2, 0

.75]

0.48

[0.3

2, 0

.62]

0.48

[0.3

2, 0

.62]

Vita

min

D(I

U)

0.58

[0.4

4, 0

.69]

0.65

[0.5

3, 0

.75]

0.51

[0,3

5, 0

.64]

0.51

[0.3

5, 0

.64]

Vita

min

E (I

U)

0.51

[0.3

6, 0

.64]

0.56

[0.4

1, 0

.68]

−0.

02 [−0.

28, 0.

26]

0.04

[−0.

27, 0

.27]

Cal

cium

(mg)

0.63

[0.5

0, 0

.73]

0.67

[0.5

5, 0

.76]

0.44

[0.2

7, 0

.58]

0.43

[0.2

6, 0

.57]

Iron

(mg)

0.50

[0.3

4, 0

.63]

0.50

[0.3

4, 0

.63]

−0.

06 [−0.

25, 0.

13]

0.65

[0.5

2, 0

.75]

Zinc

(mg)

0.51

[0.3

5, 0

.64]

0.54

[0.3

9, 0

.66]

0.49

[0.3

3, 0

.62]

0.46

[0.3

0, 0

.60]

Fola

te (m

g)0.

69 [0

.58,

0.7

8]0.

68 [0

.56,

0.7

7]0.

46 [0

.30,

0.6

0]0.

47 [0

.30,

0.6

0]


NIH


NIH


NIH


Bosco et al. IU

, Int

erna

tiona

l Uni

ts.


NIH


NIH


NIH


Bosco et al.

Tabl

e 3

Nut

rien

ts

With

out s

uppl

emen

tsW

ith su

pple

men

ts

Wei

ghte

d ka

ppa

[95%

CI]

Exa

ct a

gree

men

tbA

djac

ent a

gree

men

tcW

eigh

ted

kapp

a [9

5% C

I]E

xact

agr

eem

entb

Adj

acen

t agr

eem

entc

Ani

mal

fat (

g)0.

49 [0

.37,

0.6

0]48

%91

%

Veg

etab

le fa

t (g)

0.48

[0.3

5, 0

.61]

52%

87%

Satu

rate

d fa

t (g)

0.46

[0.3

4, 0

.58]

46%

89%

Mon

osat

urat

ed fa

t (g)

0.57

[0.4

6, 0

.68]

55%

93%

Poly

unsa

tura

ted

fat (

g)0.

27 [0

.14,

0.4

1]39

%78

%

Lino

leic

aci

d (g

)0.

31 [0

.17,

0.4

5]42

%78

%

Ole

ic a

cid

(g)

0.56

[0.4

5, 0

.66]

52%

93%

Om

ega

3-fa

tty a

cids

(g)

0.45

[0.3

1, 0

.59]

54%

84%

Prot

ein

(g)

0.51

[0.3

9, 0

.62]

51%

90%

Ani

mal

pro

tein

(g)

0.46

[0.3

4, 0

.59]

49%

86%

Cho

lest

erol

(mg)

0.44

[0.3

2, 0

.56]

45%

88%

Car

bohy

drat

e (g

)0.

49 [0

.37,

0.6

0]49

%88

%

Die

tary

fibr

e (g

)0.

41 [0

.28,

0.5

4]45

%84

%

Alc

ohol

(g)

0.40

[0.2

7, 0

.54]

48%

83%

Vita

min

A (I

U)

0.41

[0.2

8, 0

.53]

42%

86%

0.45

[0.3

3, 0

.58]

46%

87%

Ret

inol

(IU

)0.

49 [0

.35,

0.6

2]56

%84

%0.

36 [0

.23,

0.5

0]43

%80

%

Thia

min

e (m

g)0.

35 [0

.22,

0.4

8]40

%81

%0.

41 [0

.27,

0.5

5]52

%80

%

Rib

ofla

vin

(mg)

0.33

[0.1

8, 0

.47]

46%

77%

0.32

[0.1

8, 0

.47]

43%

79%

Vita

min

B6

(mg)

0.37

[0.2

5, 0

.50]

39%

85%

0.39

[0.2

6, 0

.53]

47%

82%

Vita

min

B12

(mg)

0.42

[0.2

8, 0

.56]

52%

81%

0.39

[0.2

6, 0

.53]

47%

85%

Vita

min

C (I

U)

0.53

[0.4

1, 0

.65]

54%

87%

0.59

[0.4

7, 0

.70]

59%

91%

Vita

min

D(I

U)

0.39

[0.2

5, 0

.53]

49%

81%

0.42

[0.2

8, 0

.55]

49%

85%

Vita

min

E (I

U)

0.38

[0.2

5, 0

.52]

45%

83%

0.06

[−0.

09. 0

.20]

31%

63%

Cal

cium

(mg)

0.44

[0.3

1, 0

.57]

49%

83%

0.27

[0.1

2, 0

.42]

39%

76%

Iron

(mg)

0.39

[0.2

5, 0

.53]

50%

80%

0.45

[0.3

1, 0

.59]

52%

83%

Zinc

(mg)

0.32

[0.1

8, 0

.46]

41%

79%

0.35

[0.2

0, 0

.49]

48%

75%

Fola

te (m

g)0.

45 [0

.32,

0.5

9]51

%85

%0.

34 [0

.20,

0.4

8]44

%79

%

a FFQ

1 us

ed a

s ref

eren

ce.


NIH


NIH


NIH


Bosco et al. b Ex

act a

gree

men

t for

all

nutri

ents

exc

ept p

olyu

nsat

urat

ed fa

t, vi

tam

in B

6 w

ithou

t sup

plem

ents

, vita

min

E w

ith su

pple

men

ts a

nd c

alci

um w

ith su

pple

men

ts w

ere

stat

istic

ally

sign

ifica

nt a

t α =

0.0

01.

c Adj

acen

t agr

eem

ent f

or a

ll nu

trien

ts e

xcep

t pol

yuns

atur

ated

fat,

linol

eic

acid

, vita

min

E w

ith su

pple

men

ts, c

alci

um w

ith su

pple

men

ts a

nd z

inc

with

supp

lem

ents

wer

e st

atis

tical

ly si

gnifi

cant

at α

= 0

.001

.

IU, I

nter

natio

nal U

nits

. FFQ

, foo

d fr

eque

ncy

ques

tionn

aire

.


NIH


NIH


NIH


Bosco et al.

Tabl

e 4

Supp

lem

ent

FFQ

1FF

Q2

Agr

eem

ent

n%

n%

Kap

pa [9

5% C

I]

Mul

tivita

min

s any

use

dur

ing

preg

nanc

y97

95%

9694

%0.

71 [0

.40,

1.0

0]

Ti

me

perio

d

Con

cept

ion

3938

%41

40%

0.77

[0.6

5, 0

.90]

Firs

t trim

este

r91

89%

9088

%0.

66 [0

.42.

0.8

9]

Seco

nd tr

imes

ter

9593

%95

93%

0.85

[0.6

4, 1

.00]

Third

trim

este

r91

88%

9390

%0.

80 [0

.60,

0.9

9]

Pa

ttern

Nev

er5

5%5

5%_b

Star

ting

at c

once

ptio

n39

38%

4039

%0.

73 [0

.60,

0.8

7]

Star

ting

at fi

rst t

rimes

ter

4847

%47

46%

0.75

[0.6

2, 0

.88]

Star

ting

at se

cond

trim

este

r2

2%5

5%_b

Oth

er7

7%4

4%_b

Tota

l10

198

%10

198

%0.

68 [0

.52,

0.8

5]

Vita

min

C7

7%5

5%_b

Vita

min

E3

3%1

1%_b

Folic

aci

d6

6%5

5%_b

Iron

2726

%22

22%

0.65

[0.4

8, 0

.82]

Cal

cium

1818

%12

12%

0.61

[0.3

9, 0

.83]

a Don

’t K

now

’ res

pons

es tr

eate

d as

mis

sing

val

ues.

b Not

cal

cula

ted

due

to fe

wer

than

10

mot

hers

repo

rting

use

.

FFQ

, foo

d fr

eque

ncy

ques

tionn

aire

.


reproducibility of reported nutrient intake and supplement use during a past pregnancy: a report...

Documents