RRR J Dietary Fiber and Risk of Coronary Heart Disease A Pooled Analysis of Cohort Studies Mark A. Pereira, PhD; Bilis O'Reilly, MSc; Katarina Augustsson, PhD; Gary E. Fraser, MBChB, PhD; Uri Goldbourt, PRD; Berit L. Heitmann, PhD; Goran Hallmans, MD, PhD; Paul Knekt, PhD; Simin Lia, MD, SeD: Pirjo Pietinen, DSc; Donna Spiegelman, ScD; June Stevens, MS, PRD; Jarmo Virtamo, MD; Walter C. Willett, MD; Alberto Ascheria, MD Background: Few epidemiologic studies of dietary fi ber intake and risk of coronary heart disease have com- pared fiber types (cereal, fruit, and vegetable) or in- cluded sex-specific results” The purpose ofthis study was to conduct a pooled analysis of dietary fiber and its sub- types and risk of coronary heart disease. Methods: We analyzed the original data from 10 pro- spective cohort studies from the United States and Eu- rope to estimate the association between dietary fiber in- lake and the risk of coronary heart disease. Results: Over 6 to 10 years of follow-up, 5249 incident total coronary cases and 2011 coronary deaths occurred among 91058 men and 245 186 women, Afteradjustment for demographics, body mass index, and lifestyle factors, ‘each 10-p/d increment of energy-adjusted and measure ment error-corrected total dietary fiber was associated with 414% (relative risk [RR], 0:86; 95% confidence interval [Cl], 0.78-0.96) decrease in risk ofall coronary events and 227% (RR, 0.73; 95% Cl, 0.61-087) decrease in risk of coro- nary death. For cereal, rut, and vegetable fiber intake (not «error corrected), RRs corresponding to 10-4/d increments, were 0.00 (95% Cl, 0.77-1.07), 0.84 (95% C1, 0.70-0.99) and 1.00 (05% Cl, 0.88-1.15), respectively nary events and 0.75 (05% Cl, 0.63-0.91), 0.70 (05% Cl, (0.55-0,80), and 1.00 (05% CI, 0.82-1.23). respectively, for deaths. Results were similar for men and women, Conel and fruits is inversely associated with risk of coronary heart disease. fon: Consumption of dietary fiber from cereals “Arch Intern Med. 2004;164:370-376 TETARY FIBER MAY RE- duce the risk of coro- nary heart disease (CHD) through a variety of cohort studies from the United States and Europe included in the Pooling Project of Cohort Studies on Diet and Coronary Dis- mechanisms, such as im- Author affiliations are listed at the end ofthis article. The ‘thors have no relevant Financial interest this article proving blood lipid profiles,” lowering blood pressure,'” and improving insulin sensitivity*” and fibriolytie activity" Di- ctaty liber has been found to be inversely associated with isk factors for CHD in ob- jonal studies." The association between dietary fi- ber and CHD incidence has been exam- ined in at least 10 prospective cohort std- ies? All but one" of these studies reportedan inverse association. Due to dif- ferences in methods and analytical tech- niques, the magnitude of this association for total fiber intake and for specific types of ber (cereal, fruit, vegetable, soluble and insoluble) remains unclear. Furthe more, only 4 studies!" have re- ported findings for women separately from men, Negative publication bias and re- sidual confounding by other lifestyle fac~ tors remain possibilities, We have there- fore conducted a systematic analysis of 10 Las ‘The following criteria forthe inclusion of stud {es in tis pooled analysis were applied: a pub lished prospective study with at least 150 inci dent coronary cases, assessment of usual dietary Intake, anda validation study of the diet assess- ‘ment method or a closely related instrument. ‘Through literature searches and inquiring with cexpers in the field, 14 studies were identiied that met these criteria, and investigators of 11 agreed to include their dats in the project. One study was excluded from this analysis beeause Atdlid not have data om dietary Rber intake. vestigators of 3 eligible studies, all from the ‘United states, did not agree to participate. The remining studies are described in Table ¥. The follow-up experience ofthe Nurses Health Study (NHS)"'was divided into 2 periods for analysis totake advantage ofthe repeated assessments of dlictary intake ad the long follow-up. The 1980- 1086 follow-up period is refered to as Nurses Health Study A (NHSa) and the 1086-1996 fo- low-up pestad of women who remained free of (©2004 American Medical Association, All rights reserved. ‘Downloaded From: http//jamanetwork.com/pdfaccess.ashx ?url=/data/journals/intemed/S47S/ on 06/22/2017Table 1. Characteristics ofthe Cohort Stu Coronary Disease ‘Woden energy aansoe No, o CHD casos ber nat, si Baseline Dale ot Permon-yars Age santana, tay sty Cohort” ineopion ofFallowp tacan(ango),y ola Events Deaths auestonare foal cereal Fat_vegetabie ag Men oo 107 S258, 5.08 1a 49 ar Women 12400 1077 TBO 55,25.09 % 41 Fa 109 pric Men som 1oa7 5144.65 ost a m1 42 a6 45 Women Geet iee7 = Sb100 «44-6 we 7 Fa 62 32 42 42 ‘BCH menony) 21141 10851218135 (EOTO] 1320 SBM_—Dithsoy 189 12018 AO cra Men ame 1m aso 4725.85) m2 7 ithisry 219 1801346 Women 241 105 2268340, 25-85) M2 48 Dithistony 17.7 108-21 a8 cs? Men ese tore = 14365503580) M238 inthitory 971 Women ees 1074 «= 1405503580) 4M Dithistory 143 HPAS® (mononi) 41574 108588206 S2GBT) «2st me $2 48 65 IMS (women ony) 0180 1886 20N620—— GTI) ma we 43 41 70 NASS(women nj) 81415 1980 «51201545 25.65 sr 7 Fa m0 22 3543 wo" vomeneny) 617 Tom saTOHD SES 6% sa a9 37 3658 vi Men eset 1m 3020 50,2070) m8 a ma 16 23 46 Women 1058 122872802070) m4 Fa 102 108 38 38 WHS (woman ony) 37272 1902100755 82,8€) 2 10 Fa e038 3357 Talal sao2u 200381 se aot ‘Rubens ANS, Avant Heath Sty ARI, Aerosdarais Rsk in Corman es Stuy ATBO, pa Tocopherol Bet Carsene Canc Proverion ‘us 0, od regan qusstomas FC. Fngh Modis Cine Heath Examination Survey GPS, Glostrup Poplton Sty HPFS: Hed Proleesarls Falo-p Sty, WHS ova Vlerens Heath Study; WHSa. Marae’ sth Std 108 Intention Prog WAS, Wor’ Heath Sty 985 Sb, rss Heath Sly 186-006 VP Vaserboren “Sampler xciing bein cadnasclr ass, cnce andr high raw reported clr ina ‘CHD until 1086 s referred to as Nurses’ Health Study B (NHSb). Following the underlying theay of survvaldata, blocksolperson- time in diferent periods ae statically independent, even i de- rived from the sime people” Therefore. pooling the estimates from these 2 periods i equivalent to sing a single period but takes advantage ofthe enhanced exposure assesment in 1980 compared with 1980. DIETARY ASSESSMENT Diet was measured at baseline in each study using a food fre- {quency questionnaire or diet history instrument. For the Ad- ‘entist Health Study (AHS),” only crude hiber was available for analysis. Therefore, to approximate the distribution of total di- ‘tary fiber in this cohort, we multiplied crude fiber by 3 5—the ‘atioof total to crude fiber from other studies, In addition to total lictary ber, ve examined ber intake rom 3 dilleren ood group sources, including cereals (grains) fruits, and vegetables, as well, a insoluble (hemicellulose, celluiose, and lignin) and Soluble (peetins, gums, and mucilages) fiber. Fiber [rom cereals, fruits, and vegetables was available for all studies withthe exception fof the AHS" and the Glostrup Population Study (GPS). A wide variety of foods contributed to exch fiber type with the relative contribution from certain foods varying among studies, tarchy vegetables, such as corn and peas, contributed substantially 10 vegetable iberinall studies. Only the Finnish Mobile Clinic Health Examination Survey (FMC)" and the Vasterbotien Interven- tion Program (VIP)® included potato fber with their vegetable fiber, and potato fiber was the most common form of vegetable fiber in these 2 studies. Only 6 studies (Atherosclerosis Risk in ‘Communities Study [ARIC],* Alpha-Tocopherol, Beta- (Carotene Cancer Prevention Study [ATBC|," Health Profession- als Follow-up Study [HPFS].” lowa Women's Health Study [WHS] 2° NHsb," and Women's Health Study [WHS}*) had es- limates of insoluble and soluble fiber, but there is no standard method for estimating thes fiber types based on food tables, so the results should be considered exploratory (CASE ASCERTAINMENT standardized criteria were used to ascertain cases of fatal and nonfatal acute myocardial infarction inal studies." Be- ‘cause the IWHS" had only self-reported data on incident CHD, ‘we used only fatal coronary cases from this study. We con: clucted separate analyses forall coronary events (fatal and non- fatal) and for coronary deaths. STATISTICAL ANALYSIS ‘Weexcluded approximately 1% of participants from each study if they reported energy intakes greater or less than 3 SDs from the study-specfi, log-transformed mean energy intake of the baseline popullation. Because the presence of clinical disease i slay cause detary changes, wealso excluded paricipants who reported history of cardiovascular disease, diabetes, or cancer (except nonmelanoma skin cancer) at baseline. Four studies (ARIC, EMC, GPS, and IWHS) with follow-up periods longer than 10 years were truncated to reduce heterogeneity in sty duration, Within each cohort, elativersks (RRs) (incidence rate ratios) per fiber increment were computed using proportional hazards regression models with the PROC PHREG program of SAS statistical software, version 8 The RRs were adjusted for relevant baseline demographic, lilestyle, and dietary factors. Cat ‘gories of covariates were standardized actos studies with few ‘exceptions, a follows, For disease history, information across studies included any or all of the following: sell-reported dis- (©2004 American Medical Association, All rights reserved. ‘Downloaded From: http//jamanetwork.com/pdfaccess.ashx ?url=/data/journals/intemed/S47S/ on 06/22/2017case medication use, or biologic measures (eg, blood pressure Sd serum cholesterol level). For physical sci, information fcross studies ranged from simple categories of low, moderate, and high esuresme activity a continuous metabolic index total physical activity, which wa grouped ino ques Pysi- Cal activity was unavalable for one sly, alcool intake as tinavallable for 2 studies, Thre regression models were con [ted as follows, Model 1 inca age (in yeas), energy t= take in localores pe day) smoking stats (eve, pas, ocr tent smoker and dose [I-h, Sl, 1524, and =25 cartes per iy), body mass index (2 measure of weight in kfogras di- vided by the square of height in meters, <23, 25-025, 25- £275, 375-30, or 230), physic stvity eves 13) cation (high school high school, highschool alcohol intake (0,23, 5c, 10-215, 15-30, 30-230, or =50 mLD nl ple vitamin ue (no, ys), hypercholestrolemia (no yes), and Iypertension (no, yes. Model included covariates in model 1 aindalo energy-adjusted quintiles of dietary saturated ft poly {saturated fat and cholesterol, Mode includes covarates in ‘model 2 and also energy. adjusted qunile of detay and supple- tment sourees of fle seid and vitamin “Two-sided 95% confidence intervals (Cls) were caleu- lated: We used the randoneffecte model developed by Derst- rmonian and Laird” to combine the log. RRs, the study $ecilic RRe were weighted by the inverse of the sum oftheir ‘arlances, We std for heterogeneity among studiesusing the ‘stimated between-studies valance Component statis" Before performing the repression analysis dietary Aber and allditary covariates were acjused withneach study for energy intake" We analyzed the energy-adjusted dietary fiber as acon tnuous variable (increment of 1 fd), We also examined quin- ies and deciles, based onthe cohor-specccitbutions, to de- termine ifassociatons wer linear and consistent withthe analyses ‘continuous ier Usngabecute bernie cu points we alo examined the ik of CHD throughout the fll range of fiber i take salable forall of the sties. To calculate the P vale for the test for trend across quits, patcpants were assigned the ‘median vale of thelr quintile of intake, and ths variable was tered a a continuous term in the Cox regression model, Re- ful of dca fiber as continuous vrable were corecte for thas due to detary measurement ero, n ber ony, using the regresion callraton method." Ths correction could nou be performed for ber intake from speciicood sources, becatse ew {tthe validation studies included these sources of Mb. Mes urement error corection was not perforied on other covai- ics and dietary factors in the model ‘Weert whether thefllowing variables mifed the association between br inlake and isk of CHD: sex, age (10- yea eategores)fllow-up time, body mass index (235, 95-30, S30), cigarette smoking (never smoker vs former or curent Stoker, saturated fat intake (percentage of energy intake q- ts), and history of hypertension and kyperholetercemia (post lve venegatvo) For cach lator of incest across predict er ofthe score for the level ofeach factor ad intake of fiber ex- Dresed aes continuo vaiable as included in septate mul tvariae models, The pooled P value for the test for elfect mod feation wae obtained using squared Wald statistics by pooling thestuy spec intecioncouTicentsanddvdingby te square ofthe SE ofthe pooled interaction term and refering thers ing statistics toa x dstrbution with. The ck of any statie- tay sent mation by ae ofl esp ports the assumption of proportional ards ss} A otal of 91058 men and 245 186 women, contributing 2500581 person-years of follow-up, were included in these analyses, The total numberof events was 5249, cluding 2011 fatal cases (Table 1). The media bt in tales for each cohort are given in Table L The RRs ofall major coronary events (ltl and non- fatal) and coronary deaths for each 1O-/d increment of cnergy-adjsted tual dietary fiber inake ate given in Fable 2. In analyses adjusted forall demographic and nondletary style factors, foreach 10-g increment in Atcary fiber, we observed pooled feduetions in risk of 1296 forall coronary events and 19% for coronary deaths, There waste attenuation ofthese pool estimates with further adjustment for dietary intake of aly acids, cho- lesterol (model 2), and dietary and supplemental folic acid and vitamin E(omodel 3). These assoeations were stale Tar for men (RR of coronary death, 0.82; 95% Cl, 0.72- 0.94) and women (RR of eoronary death, 0.80; 95% Cl 0.66-0.96) Further adjustment for alpha and beta caro tene; 3 maine laty acs, andfinolenic aid did nox materially change the results (data not shown) Analy- sls of dietary fiber quintile revealed simular ndings (RR ‘trop quintile compared with the bottom quintile 0.90 Tora events; P< 09, tet for end: RR for coronary deaths 0170,P<.001, test for wend). The results for model 3 were ‘corrected o las due lo measurement error in ber only the RRs associated with L0og/d increment were 0.86 (951% C1, 078-096) forall coronary events and 0.73 (5% C1, 0.61-0.87) for coronary deaths he results for types of berate summarized in '3, with adjustment for all demographic, e- style and dietary factors as we dd for model 3. We ob- Zerved pooled reductions in risk of all coronary events ot Lom foreach l-gid increment of cereal an 1O% pet To-g/d increment of fruit ber although the finding for cereal iherhad al that included 1.00, Asociatons were tonger for coronary deaths thas for allevent, with dlctions in sk of 298 for crea ber and 30% for fruit Aber foreach 10 increment. n contest, vegetable fi ber was not asotiaed with CHD Incidence or mortl- My, Heterogeneity (P=.025) was observed in RRs among the @ studies included in the analysis f cereal fiber and all coronary events. This heterogeneity seemed tobe ex- plained bys scx dlerence due to postive associations tn cohorts of women—ARIC, NHSa, and VIP. No sige nificant heterogencity was observed tn any of the ether analyse “To determine if the associations observed for cereal independent, we included these ber types in the same fegression model The resus ofthe shulyss were similar forall events (ult ibe: RR, O81; 95% Cl 0.69-0.95; cereal fiber: RR, 0.89; 95% CI, 0.7 1.05) and deaths (uit fiber: RR, 0.65; 95% Cl, 0.49- 0.8; cereal fiber: RR, 0.71; 95% Cl, 0.59-0.87), suggest ingihat the ects crcl andr ber ere indepen leach other. We also exatined the asocialions be- tween soluble and insoluble fer and CHD risk. Intake ofboth iypes of fiber was inversely associated with eco all coronary cents and of coronal deaths, Noheierog nity was obecved among the Rs The asocations we Stronger for soluble ber all events: RR per 10-g/ ine ment, 0.72; 05% CI, 0.55095; deaths: RR, 046; 95% CL 6128-074) than for insoluble fiber all events: RR, 0.90, 95% Cl, 0.83-0.97; deaths: RR, 0.80, 95% C and fruit fiber wer (©2004 American Medical Association, All rights reserved. ‘Downloaded From: http//jamanetwork.com/pdfaccess.ashx ?url=/data/journals/intemed/S47S/ on 06/22/2017Table 2, Study-Speetic and Pooled Multivariate Relative Risks (954 Confldence Intervals) ‘of Coronary Heart Disease per 10-g/d Increments of Total Dietary Fiber ‘Alene Deaths as ‘Men on os tonaasasty 05s 02 032 (005218) Women 130 153 TaN(OM6-435) 185 145 185(0.28121) alc Men oo too tare 070 on as4aro457) Women 068 067 = o7g(oa043) © a7 ‘BC (men ony) 09% 055 005 085-109) ae os ose(ase-.aa) AIC Men on oso © og7a7e129 © 08s 73 o7eiouot2) Women ost Ose omaiaaz43t) 088 ou 033 (000122) ors ‘Men oar oss osg0se1sy 100 1531.86 086.420) Women 192 tus 100038298 © os 037 (004391) PES (nen ony) 085 ost = oga7eos7) §=— ar? 0s 0.84 (060-103) [MHS (wore ai) 078 078 sn (o6t-t08) Sa (women ol) 095 054 ogra 105 114 a6 055-167) Sb (women an) oa og = ogr(a72108) 79 os 075 (058-107) ve Men 095 ose = otseis1y ogo 05053 (02t415) Women at zo 818(098-037) 0s Was (omen ony) 075 07s orn @so447) 18 410177 ;024-1290) Poolet 0.98 (084089) 0.8 (089-0.98) 0.81 (080097) 0.81 (0749.88) 0.81 (0.7-0.92) 0.81 (073-081) P al for alate ri <2001 <001 08 001 m1 oot Pal for htrognay 2 30 2 5 20 73 ‘Mer metsursment arr eorcton 185 (07.096) 073 081-087) Pa for alee 005 oot Pal for hatrogenty 28 ry ‘Abrevtons: AVS, Advent Heath Stuy AIC, Atbroslose Rsk n Corsmnies Study, ATR, Alhs-Tocaphero, Bat Carlee Cancer Prevanton Stay A, Finish bie Cin Hath Exarination Suey: GP, Glostup Population Susy HFS, Heath Professions Folow-up Stud NHS lea Womens Health Sty MHS, res Hanh Study 180-1085, AS, Muses Hath Sts) 1085-1086; VI, Vasterboten trenton Program: WAS, Ware's Heath St liga, tant appa “Hdl alae as (nyas), nea nak nko pr day). sain status (nee, past cunt maker and os [1-4 514, 15.24 and =25 cigars per day), buy mas fox (eld as weight in lores ve bythe square of height nmr 23, 23-<25,25-<27 5,27 8-30. =20), psa! tbiy (evel 13), duet (=high choo! high school igh schoo), alehol ake (0, <5, 5-10, 10-18, 1-30, 3-80, 0 mL), mule Suma s(n. ys). petcholssalemi snd ypararsion Hod! 2 des cavaratas node 1a ale egy asta guns of tary sna Polyunsaturated at archos. Coniene itera ana shown fot mods ard? ar snp of pesaraton ode inclues covariates in model 2 and $a energy ssed gut of datay snd supplement saues of fae dan ami Both ARC women and VP women a nse eases fo nals eats “abet frm association. Teva tet or bean tds hterogeniy 0.69-0.92). The results were similar, although with wider coronary mortality (27% reduction in risk for each Cis, when soluble and insoluble fiber were included in. 10-g/d increment in total dietary fiber) than for all the same model. events (14% reduction in risk). Although cereal and The results for measurement error-corrected total__fruit fiber had strong inverse associations with CHD dietary fiber were generally consistent when stratified by risk, no stich associations were observed for vegetable age, follow-up time (including exclusion ofthe irst2years fiber. These associations seemed to be independent of or stratification by first and second 5-year periods), over- other dietary factors, sex, age, baseline body mass weight status, smoking, and intake of saturated fat data index, smoking, history of hypertension, diabetes, and not shown). There were no significant interactions be-_hypercholesterolemia. tween dietary fiber and any of these variables. The RRs were generally consistent across the stud- Finally, the results according to absolute cut points ies. The only observation of heterogeneity in RRs was for of energy-adjusted dietary fiber are presented in the: the analysis of cereal fiber and total coronary events, in wre for coronary mortality. The reference category was which 3 cohorts of women (AHS, NHSa, and VIP) had 18 to 21 g/d and adjustment is the same as in Table 3, RRs greater than 1,00. In NHSs, an older version of the food frequency questionnaire was used, with limited in- f comment formation available for quantifying total fiber and espe- cially cereal fiber. The relative contribution of refined. The results of the present study suggest that dietary grains to cereal fiber in NHSa seems to have been exag- liber is inversely associated with risk of CHD in both __gerated, whereas the opposite seems to have occurred for men and women. The associations were stronger for whole grains. Because whole grains, but not refined grains, (©2004 American Medical Association, All rights reserved. ‘Downloaded From: http//jamanetwork.com/pdfaccess.ashx ?url=/data/journals/intemed/S47S/ on 06/22/2017Table 3. Study-Specitic and Pooled Multivariate Relative Risks (95% Cont ) of Coronary Heart Disease per 10-g/d Increments of Cereal Fiber, Fruit Fiber, and Vegetable Fiber oo one Seon rat wpa ime in 08 om tg om om im in 1s Fe 1g smo) ie on ost sai oa Ea fi in oss ta to ow 1 oat im os ie on a tS tn ws nei on on a a un te ts oer oe Fa oa me come) ita Fa iti ts te % eon) Ee tae is aa te BS we in on tm a ost ove an tm aS ia on em FS 0 om ois as sii ro ssoenio emanem mest anemam omens 1mesiz Paterna a . as an = Potato a « = = : ‘Abbreviations: ARIG,Aterosclrosi ik in Communes Stay ABO, Alph-Tocopheral, et Caotne Cancer Provan Study, FMO,Fanish Mobile Cnc Heath Examination Survey, HPA, Heath Protesiarals Folawap Stuy HS, iowa Womens Health Study, NHS, Narsos’ Heath Study 060-1085; HAS, Nutss Heath Study 198-1006, VIP, Vastrbten Irtrvenion Program: WHS, Wore Hal Stuy ellpss, tana appeabe, “adel cludes age (Jar ae groups), energy aka (nKlocaaros pr day), smokng aus (nvr, past of current sor and dose [1-4 5-14, 1-24, and =25cgaratie per day), boa mass indo (lela as weightnklograms dia by he square of hoight in metas] 23, 23-<25, 25-27 8,27 8-<20, 0), phys ari (evel 1-5), edueaton (highschool. igh schoo, "hgh schoo, aleahol make (0,9, 5-—10,10-< 16, 16-<30, 30-80, 30 mL), ‘multiple vari se (oye), istry of hypercolstaolnia and hypartnsonandaeiy-ajustd quires fda satura fa, ojunsatrtd tt and Cholera and quite fdr nd cupplement saves o fle aed and warn €. Bo AIC women and VIP women ha isuicent case or anal of ‘hte "Pau, et fr no association. Teva tet or bean tds hterogeniy Four ofthe studies findings on fiber and CHD tn- cluded in the poling project had been previously pub lished" Inthe NES andthe HFS, te suomest ie verse assocttions were observed fr cereal ibe, With ~. t Weaker associations for fruit and vegetable fiber. in the f TBC verse assolalons were generally observed for flltypes of ber. Inthe WHS, Lit etal” observed the Strongest suselions or fut iberlniake and ck of otal Carleocula dlcae, whereat be asocsions Were ch- Scrve wth incidence of myocar infarction slp ished studleson tberand cliD were no eluded in the pong project beertae they did ot meet there Troms sa least 150 incident cases" prac oval Rabe aD Om Te Til stay Fra, Ee ted eta isk of deat fom coronary heart disease (CHD) by category octal Sitay Ser nak. The rove ease ast ora same erates se dated dietary assessment” or we were previously un- inal Er bars net cones an fare of their existence." Of thes, 3 reported stalisticaly Significant inverse associations bewween dietary fiber in- have been shown to reduce risk of CHD." such mes-_takeand CHD," 2 reported inverse associations that surement error in cereal fiber intake could explain the were not statistically significant," and 1 study re- tinexpected NHSa findings. Indeed, the previously pub- ported a nonsignificant positive asociation.* Although lshed findings of NHS included analysis of the dietary Mann etal observed a nonsignificant increased risk of Aber intake average over the repeated food frequency ques CHD with increasing total fiber consumption, this find- tiomnaites (1984, 1986, and 1990) and the results re- ing may have been spurious duc to the small number of vealed a strong inverse association —an RR of 0.63 for events (38 deaths) each 5-yid increment in cereal fiber." The findings for There has been tte suppor for an inverse associa- the women ofthe AHS and VIP were not consistent with ton between vegetable iberitake and risk of CHD. One those for the men in those studies. Furthermore, be- possible explanation for this finding ls the nutrient- cause the Cls were very wide for these estimates, we ate poor high glycemic load nature of common starchy and unable to draw any meaningful inferences from them. heavily processed vegetables, such as corn and peas, Two (©2004 American Medical Association, All rights reserved. ‘Downloaded From: http//jamanetwork.com/pdfaccess.ashx ?url=/data/journals/intemed/S47S/ on 06/22/2017studies (VIP and FMC) also included potato in their veg- cuable fiber analysis, Dietary glycemic load has been shown to substantially inerease risk of CHD and type 2 diabe- tes mellitus."“* Therefore, any beneficial effects of veg~ table ber may be countered by some adverse effects of starchy vegetables. More attention needs tobe given, both in research and public health recommendations, to the types of foods being studied and recommended. Assuch, one limitation ofthe present study isthe absence of food dlatato complement these analyses on fiber. Although such pooled analyses of foods and dietary patterns in relation to CHD are beyond the scope of the current investiga- tion, they should be ineluded in future efforts. (Of additional interest is whether protection from CHD may come from both soluble and insoluble fiber. Previous studies*™* have supported this possibility, with no consistent advantage of either class of fiber. Al- though we observed inverse associations for both types of fiber in the present analyses, the RRs were stronger for soluble fiber, reaching 0.46 for risk of coronary mor- tality per each 10-9/d increment. These results must be interpreted with caution, because only 6 studies est mated insoluble and soluble fiber, and there is no stan~ dard method used to derive these estimates. However, & characteristic of soluble fiber that may explain these find- ings is its propensity to increase intraluminal viscosity of the small intestine, thereby slowing the absorption of nutrients and potentially binding bile acids.” Such ef- fects have been shown to decrease insulin seeretion and improve glucose control," lower serum cholesterol lev- cls," and possibly lower blood pressure."® Neverthe- less, the finding of inverse associations between both soluble and insoluble fiber and CHD risk in the present analysis supports recommendations to increase con- sumption ofall ypes of fiber-rich foods. ‘An advantage of the pooling project is the inelu- sion of previously unpublished results that may have been susceptible to negative publication bias inthe past. Thus, the pooled results may be closer to the true asso~ ciation than individually published findings. Other advantages include the systematic conduct of the ana- lytic strategy across all studies, modeling exposures and important covariates uniformly. Such efforts decrease the likelihood of heterogeneity among RR estimates, thus enhancing generalizability of the pooled estimates. Therefore, the pooling project makes the best use of the available observational data to address hypotheses about diet and chronic disease. Although the ability to use the data from validation studies to correct the dietary fiber for measurement error was a strength of this analysts, the measurement error correction must be interpreted with caution because we were unable to adjust all of the covariates and other dietary factors for measurement error. Other limitations include the het- erogeneity of dietary assessment and food table meth- cds. For the soluble and insoluble fiber analyses, in par- ticular, there is no accepted method of measurement, and only 6 of the studies had quantified these fiber types. However, we found only one instance of statisti- cally significant heterogeneity in the RR estimates among studies, suggesting that the limitations of our methods did not undermine the validity of the findings. In conclusion, our results suggest that dietary ber intake during adulthood is inversely associated with CHD risk. Coronary risk was 10% to 30% lower for each 10- 1d increment of total, cereal, of fruit fiber. These re- sults provide strong confirmation of the results of pr viously published cohort studies, nd they are supported bby numerous experimental studies that demonstrate a ‘wide range of possible biological mechanisms through which liber may reduce the risk of CHD. Therefore, the recommendations to consume a diet that includes an abundance of fiber-rich foods to prevent CHD are based, ‘on a wealth of consistent scientific evidence. ‘Accepted for publication March 12, 2003. From the Division of Epidemiology, School of Public Health, University of Minnesota, Minneapolis (Dr Pereira); Departments of Nutrition (Ms O'Reilly and Drs Willett and Ascherio), Epidemiology (Drs Spiegelman, Willet, and Asck- rig), and Biostatistics (Dr Spiegelman), Harvard School Public Health, Harvard Center for Cancer Prevention (Dr Willet), Channing Laboratory, Department of Medicine (Dr Willet), and Division of Preventive Medicine (Dr Li), Brigham and Women’s Hospital and Harvard Medical School, Boston, Mass; Department of Medical Epidemiology, Ka- rolinska Insitute, Stockholm, Sweden (Dr Augustsson); Cen- ter for Health Research, Loma Linda University School of Medicine, Loma Linda, Calif (Dr Fraser); Section of Epi- demiology and Biostatistics, Henry N. Neufeld Cardiac Re- search Institute, Sheba Medical Center, Tel Hashomer, Is- rael (Dr Goldbourt); Institute of Preventive Medicine, Copenhagan University Hospital, Copenhagan, Denmark (Dr Heitmann); Department of Public Health and Clinical Medi- «ine, Umea University, Umea, Sweden (Dr Hallmans); N. tional Public Health Institut, Helsinki, Finland (Drs Knekt, Pietinen, and Virtamo); and Departments of Nutrition and Epidemiology, School of Public Health, University of North Carolina, Chapel Hill (Dr Stevens). This work was supported by research grant ROL HL58904 from the National Heart, Lung, and Blood Insti- tute, National Institutes of Health, Bethesda, Md. The GPS was financed by the FREJA (Female Researchers in Joint Action) programme from the Danish Medical Research Council We thank Karen Corsano, LMS, for computer support ‘and Julie E. Buring, SeD, David Hunter, MB, BS, Stephanie ‘Smith Warner, PhD, Shiaw Shyuan Yaun, MPH, and John Ritz, PhD, for their expert advice. The steering committee {for the Research Unit for Dietary Studies and the Copen- hhagen County Centre for Preventive Medicine made data ‘available from the GPS. Corresponding author: Mark A. Pereira, PhD, Divi- sion of Epidemiology, School of Public Health, University ‘of Minnesota, 13008 Second St, Suite 300, Minneapolis, MN 55454 (e-mail: pereira@epi.umn.edu). Ls} 1. ein QUA, Ken CWE kan Va Sle fe ina at dose op ody th US aod and rg station fac of heath ene: ‘eran i kat for earonscur dese asesedin aranéomaed cor ‘led roseover al Am J ln 20027 3-808, (©2004 American Medical Association, All rights reserved. ‘Downloaded From: http//jamanetwork.com/pdfaccess.ashx ?url=/data/journals/intemed/S47S/ on 06/22/20170 0. w * u 6 % m. 0 Jes DA, Newton A, Leeds AR, Curmings J ec tp, guar gu, {Sd eat eon env choles. Lancet 75111617 ‘bent Ancrson SU, ela MP, a Pum nich cease aed taal else and LDL colesro but not HL chaste yperchas ‘alam aus: esas om arta. J Mu 1671271873100 nan JM Ps Fra, Moran Turgut. Ot ingestion ees 5 ta ard dasa bad pressure among moda yperensies2 piet Sram Prac 200251298, ‘alan V Jenkin 0, pdr Pt Kaien (lacomsnan in rove glyems a other ae ik tare or coronary a deseo ‘ype 2 dabews: 2 anamaed cooled maths ia Dabs Ca 1086 zzoieo10 Foukagava DK, Anderson JW, Hageman 6, Young VR Miaker KL. High carbo ight cas icrease parighral msl sensi in hay Young an lads Ar. Cin ur 190052528 28 Prenat DJ, Ps. tal Te fet owl grains on inl ans in veri hyperisulnenic dus A.J i 202 75348 as Para MA Pin. ita ter and carer Sas: xeric and ‘pidge stenes, Cr Atal Rep. 20002404 512 Humble CB Marcher lk, Tyo Diary be and ornay hart de aan mieagedtyparthlestleric men. A J Pov Me. 19859107 me Lig DS, Pereira MA, Kren Ht Diary er wht ana oat las iene rik acters in young adults AR 1052821590156, Duss lian RC, ag Y Are DK, Stns Boe Relationship vee dtr ar conuntion and atin nd pimagen acorn "tor pe te aol ear Lung and oe ste Fay Hea Sty. ‘rnd Cin at 198 8568-75. Feshos EI Lace, KemoutD. Dian py ait 2s etermiats ftps the Zhen ety Stuy. Am J pel 0010.50 ‘xo, iv, ing J, Sees, im Wet W Mason J. prospec ty ot layer aa rie ot earonselr dese amg women JA Call (att 200230406 Wok Marcon ME, tpt Me Longer it of itary Ser and ‘erated kot oon hea se among amen ANA 10928199 ae, Fie 8, AscoriA GixanucE,SigenanO, Stumpf MU We WO ‘age, ut anda ar ts nico ornay has diese rang nen AA HOG 275447451 Fete , nm 3 Koronen Peake feather and rik a co ary heart dias in a oho of Fn mat: th AlhaTocopeol, Bt (aon Cara Prevention Stal. Cveabton10959827202727 Mots JN. Mar JW, Clayton DS, Dt and ear: posts. UL 19772: nora ‘Mana. Apply Pl, Key, Therogo MDa detrinats at ischas est eae nbesth costae nd. Aewr TOT TED 58, aa Wonovard, Tune edo, Baio Sth Dietary ais ‘us and therin thal of carovascuar ease ard a ase ora ‘euro te Sein Han Heath Study. Am J Epo. 10050 1073-080 ‘un KT, aret-Conor Day fran auc scenic hat dase ‘moa asin men até wore: a 12,erpospectie sy. An pe mol 1087 12600-1102 "shit Lew RA Sa, it an 20 jar orto coronary aad ase th reaped Boston Det Hen Sud MEd 108531281 18 asa GE Soba J Beeson, Satan TM ApossBlpatei eet at (gepniyrep) SRGHINTERN WEDVOL or FER ao a. cansurpion ania coreay a Ine 092.952 1416-128 rfot 0 Latsan 6 von de Lit, eho M. Hoty, dnc of ete ‘yo ircn,andmeralyinaampe ofan an nd warn ‘ins Ede 1085127748, Kit aunenen A Janne, Sopinen , Hive M, Arama A Ar ‘ar amin aad coro may inant option st, ‘nS Eder 1984 19:80, ‘Wot Heath Orgenzain WONIA Poe Prinpl nestiaors. Toe Wort Hath ganzatr MONCA Po tring rads nd termina incr toascla diese arjer trator clabraon. Clin Eien 185 a5 114 Falta AR, met OX, Hutchinson RG, Lt Clog LX, Cope LS. Physical ‘vy ardent orarar hea easel aged women nen Med Si Spons Eee. 195720901900 sit Flom AR Pie, Mk, WY, otek RM Dir an atari ad daz rom coronary aa dseasin posmenapasal woren, Eng Je, 1006 33412-1162 esha Advent Heth Sty Arch 2% Rohan Greenland. Maden poison a, Pasi, Pa: Lip pico Wins & Wis; 198 28. SASISTAT Sotae: The PHREG Proce Primi Oscmetaton iy, CSAS It In 1991, 31. DeSemnian La. Meas inci ils Coorl Cn is. 186; a 2 Tara Wie We. tuatonl Epsemalogy 2 ed New Yor, MY Ona Unversiy Pres 1008298 201, ose, it WO, Spiegel D Corscton of gsc rpesson lve iestinaes and contioes tars for tana wt petson muse entero St Med 198 81051-1068, Spielman. McDemet A, Rosner 8. Rosin calbrason maths fren ‘etna bas nina episrilg.A JCn - 107 eri. 1a6s, aos OR Hyer MA Kashi Folsom AR. Whole grin intake may ee ise coronary at diese eth postop waren: tela Wan ‘ets oath Sy Ar Co Mr 10865828 357 iu, Stmper Ml, Hu etal. Wole-gan consumption an rik of oxanry lease es ram th sa alt Sy Arm Cin Ma 10070 aig Liu, Wien Wo, Stampfer My t aL A prospective sy of aay gheemic log carolyn and ik of nanny aa ease in US wom. Am in ar 200071485-46. Salman J Ascari Rem 8 eal Dery Sb, emia. and kot OM in men Dates Cre 187 20585 50. Salmaon J Manse J, Stamper AU, ta Distr er, conic ad, and "i of no-isu-éopendent dabts mals in women, ANA 1007277 aa, 38 Marie. Hs, Von, Stanek Hand, Sty A Mehr rf ram ctl educa ot bran, Hapa D420 M5657. 14. raat J, S20 FH, Wood Pl High bo-gacs ot anon gum edie 4 2 estranda tae coe nds apt with and what ype 2 dia ete aot Md 1006721208 Chana Garg Luann, Bargmann KY, Grundy SM ine Len ‘alti of igh tay ther mgt wih yp 2 dbs ml. ‘us Fag Me, 200032 1302-1308, Laenen Pane iS, Mydanen HM ye bead aes sunt and \oLetlsteolinmen uth madre evae serum hos, Ja 200 ‘as6e-70 (©2004 American Medical Association, All rights reserved. ‘Downloaded From: http//jamanetwork.com/pdfaccess.ashx ?url=/data/journals/intemed/S47S/ on 06/22/2017