Dairy products and health: a review of the epidemiological evidence

Dai y p oduc s and heal h: a e iew o he epidemiological e idence E a-Elisa A ´l a ez-Leo ´n 1 , Blanca Roma ´n-Vin ˜as 2 and Lluı ´s Se a-Majem 1,2,3 * 1 P e en i e Medicine Se ice, Complejo Hospi ala io Ma e no-Insula , Las Palmas de G an Cana ia, Spain 2 Communi y Nu i ion Resea ch Cen e, Scien i ic Pa k o Ba celona, Spain 3 Depa amen o de Ciencias Clı ´nicas, Uni e sidad de Las Palmas de G an Cana ia, Ap o. Co eos 550, E-35080 Las Palmas de G an Cana ia, Spain E idence-based nu i ion is essen ial o mo e o wa d in he science o communi y nu i ion. The p esen s udy is a e iew o he epidemiological e idence o dai y p oduc s and heal h. The e is an in e se associa ion be ween he in ake o dai y p oduc s and hype ension, s oke and colo ec al cance . The e is no e idence o an associa ion be ween he consump ion o dai y p oduc s and b eas cance . The e is some e idence linking high- a dai y p oduc s and an inc emen al isk o p os a e cance and weak e idence o he p o ec i e capaci y o dai y p oduc s on bone heal h. Mo e p ospec i e s udies should be de eloped in o de o es ablish be e e idence o he ela ionship be ween dai y p oduc s and heal h. Due o he impo ance o dai y p oduc s in public heal h nu i ion, quan i a i e ecommenda ions should be es ablished in he ligh o he scien i ic e idence. Dai y p oduc s: Public heal h: Re iew li e a u e Public heal h nu i ion should no be unawa e o he impo - ance o e idence-based medicine. Sys ema ic e iews o he epidemiological e idence and he c i ical analyses o ha e idence ha e esul ed in a huge ad ance in he de elopmen o p o ocols and guides o clinical p ac ice. E idence-based medicine has esul ed in mo e app op ia e ac ions o sol e e e yday clinical p oblems, based on he c i ical e alua ion o such e idence, ins ead o in ui ion o expe ience. To de elop public heal h nu i ion policies based on scien- i ic e idence, i is impo an o igh agains misin o ma ion o he gene al popula ion abou nu i ional aspec s o hei daily ood in ake. F equen ly, he bene i s and isks a e pu down o dai y p oduc s on he basis o physiopa hological heo ies, bu wi h no epidemiological s udies o suppo hem. The objec i e o he p esen wo k is o e iew pape s co e - ing dai y p oduc s in bo h heal h p omo ion and disease p e- en ion. Focus is placed on he c i ical analysis o he scien i ic li e a u e, poin ing ou u u e needs in he esea ch o e idence be ween dai y p oduc s and heal h. Me hods Rele an a icles we e ob ained h ough sea ching he MED- LINE da abase ( om 1966 o Janua y 2005). Keywo ds used in his sea ch we e he MeSH e m ‘dai y p oduc s’, de ined as ‘ aw and p ocessed o manu ac u ed milk and milk-de i ed p o- duc s’ ha included he ollowing p oduc s: bu e , cheese, ice c eam, ma ga ine, milk and cul u ed milk p oduc s (yoghu ) and some heal h condi ions. The selec ion o heal h condi ions was ca ied ou on he basis o hei public heal h impo ance (p e alence and/o se e i y) and included neoplasms, CVD (myoca dial in a c ion, hype ension o ce eb o ascula acci- den ), and os eopo osis o bones. The MEDLINE sea ch was complemen ed wi h he au oma ic PubMed clinical que ies s a - egy, which combined ‘dai y p oduc s’ sea ch wi h ci a ions iden i ied as sys ema ic e iews, me a-analyses, e iews o clinical ials, e idence-based medicine, consensus de elop- men con e ences and guidelines. Ci a ions om jou nals spe- cializing in clinical e iew s udies we e also included. To be eligible, wo ks had o ul il he ollowing c i e ia: abs ac mus be a ailable, be w i en in Spanish, English o F ench and co e human popula ion. P io i y was gi en o me a-anal- ysis and sys ema ic e iews. E idence om epidemiological s udies o dai y p oduc s and heal h was summa ized. To e alua e he s eng h o he e idence, alidi y and accu acy we e aken in o accoun . Validi y was assessed by analysing he p esence o possible sys ema ic bias in he design o execu ion o he s udies. Accu- acy was assessed by conside ing he size o he sample and he wideness o he con idence in e als. Resul s The e we e mo e han 85 000 a icles on he consump ion o dai y p oduc s. Once he inclusion c i e ia we e applied, ou - een me a-analyses o sys ema ic e iews abou dai y p oduc s and he selec ed heal h condi ions we e iden i ied, six on cance , six on CVD and wo abou bone heal h. The esul s a e p esen ed sepa a ely by heal h condi ion. *Co esponding au ho : P o esso Lluı ´s Se a-Majem, ax þ34 928 453475, email [email p o ec ed] Abb e ia ions: DASH, Die a y App oach o S op Hype ension; DBP, dias olic blood p essu e; OR, odds a io; RR, ela i e isk; SBP, sys olic blood p essu e. B i ish Jou nal o Nu i ion (2006), 96, Suppl. 1, S94–S99 DOI: 10.1079/BJN20061709 qThe Au ho s 2006 h ps://www.camb idge.o g/co e/ e ms. h ps://doi.o g/10.1079/BJN20061709 Downloaded om h ps://www.camb idge.o g/co e. ULPGC. Biblio eca Uni e si a ia, on 24 No 2017 a 09:33:19, subjec o he Camb idge Co e e ms o use, a ailable a Dai y p oduc s and cance The e we e six me a-analyses and one sys ema ic e iew ha speci ically analysed he consump ion o dai y p oduc s and cance isk (one o colo ec al cance , h ee o b eas cance and h ee o p os a e cance ). Colo ec al cance One o he a icles analysed he isk o colo ec al cance . I was a pooled analysis o en coho s udies ca ied ou by Cho e al. (2004) as a pa o he Pooling P ojec o P ospec i e S udies o Die and Cance . This pooling p ojec accumula es da a om sub- jec s who had pa icipa ed in di e en p ospec i e s udies wo ld- wide. Once amassed, he da a we e eanalysed, which esul ed in a bigge sample size and longe ollow-up pe iod han hose o he o iginal s udy alone. To da e, h ee pooled analyses ha e been published o his pooling p ojec : a in ake and b eas cance ; alcohol in ake and colo ec al cance ; and mo e ecen ly, dai y oods and Ca and colo ec al cance . Fo he pooled analysis o dai y p oduc s and Ca and colo - ec al cance (Cho e al. 2004), en p ospec i e s udies we e iden i ied. S udies had o mee p ede ined c i e ia: assessmen o die a y in ake by a alida ed me hod, and a leas i y people diagnosed wi h colo ec al cance . Finally, 534 536 men and women we e included, wi h a ollow-up pe iod ha anged om 6 o 16 yea s. In ha pe iod 4992 inciden cases o colo ec al cance we e documen ed. In ela ion o consump ion, esea che s ook in o accoun di e en g oups o dai y oods: milk (including skimmed, low- a , medium- a , whole, e apo a ed and bu e milk), cheese (including high- a , ha d, co age and ico a) and yoghu (including low- a , egula yoghu and yoghu d essing). They also conside ed in akes o Ca om he die (and om supplemen s when a ailable) ca ego ized in deciles. O he a iables associa ed wi h colo ec al cance isk we e also conside ed. Resul s we e adjus ed o heal h- ela ed a iables BMI, heigh , amily his o y o colo ec al cance , use o non-s e oidal an i-in lamma o y d ugs and use o mul i i a- mins), li e-s yle a iables (smoking, educa ion, physical ac i i y, alcohol in ake) and die a y a iables (in akes o o al ene gy, ed mea and die a y ola es). In women, o al con acep- i es and pos menopausal ho mone use we e also conside ed. O iginal s udies p o iding subjec s o he pooling p ojec came om he USA, Canada and no h Eu opean coun ies (Holland and Sweden). This geog aphical a iabili y ga e he p ojec a wide ange o in a-indi idual a ia ion in he consump ion o dai y p oduc s. Mean in ake o milk anged om 137 o 687 g/d, mean in ake o yoghu anged om 11 o 104 g/d, and mean in ake o cheese anged om 8 o 27 g/d. Die a y Ca in akes anged om 674 o 1051 mg/d. Milk consump ion was in e sely ela ed o colo ec al cance . Compa ed wi h pa icipan s who consumed less han 70 g o milk daily, he ela i e isk (RR) o colo ec al cance was 0·85 (95 % CI 0·78, 0·94) o hose who consumed 250 g/d o mo e, as shown in Fig. 1. In o he wo ds, people who consumed less han a qua e o a glass o milk e e y day had 15 % mo e isk o colo ec al cance han hose who consumed $1 glass o milk a day. The associa ion be ween milk consump ion and educed isk o colo ec al cance was limi ed o cance o he dis al colon and ec um. Cheese in ake was analysed in wo ways. Subjec s who consumed $25 g o co age o ico a cheese daily had a RR o 0·83 (95 % CI 0·72, 0·96) o colo ec al cance , compa ed wi h hose who did no consume co age o ico a cheese. Consump- ion o o he kinds o cheese – excluding co age and ico a – was weakly posi i ely associa ed wi h colo ec al cance , al hough con idence in e als we e no signi ican o any o he di e en ca ego ies no was he es o end be ween ca ego ies. Yoghu in ake showed a weakly p o ec i e p ope y o e colo ec al cance , al hough CI and es o ends we e again no signi ican . O he e men ed dai y p oduc s (including yoghu , bu e milk and sou c eam) showed simila esul s o hose o yoghu . The o he dai y p oduc s measu ed, including bu e , c eam and ice c eams, we e analysed, bu he e was no s a is ical signi icance wi h ela ion o colo ec al cance isk. As shown in Fig. 1, a high in ake o Ca was also associa ed wi h a lowe isk o colo ec al cance . Subjec s in he highes quin ile o die a y Ca in ake had RR o 0·86 (95 % CI 0·78, 0·95) o colo ec al cance , compa ed wi h subjec s in he lowes quin ile. Fo o al Ca (combining die a y and sup- plemen al sou ces), he RR o he highes quin ile was 0·78 (95 % CI 0·69, 0·88). In conclusion, milk and Ca in ake we e associa ed wi h a lowe isk o colo ec al cance . One o he s eng hs o he pooled analysis p esen ed he e is ha all s udies included we e p ospec i e, a oiding he bias associa ed wi h he di e - en ecall o die a y habi s once a subjec knows he diagnosis o a diges i e cance . Ano he s eng h is ha all s udies used alida ed die assessmen me hods, minimizing he possibili y o an inco ec egis a ion o he ac ual in ake o dai y p o- duc s. Finally, he analysis o a da abase o mo e han 500 000 subjec s om di e en coun ies ga e su icien s a - is ical powe o analyse speci ic dai y p oduc s ca ego ies. Howe e , i is ema kable ha he au ho s did no adjus o some die a y a iables ela ed wi h colo ec al cance , such as ib e o ui and ege ables. Al hough he au ho s men- ioned ha addi ional adjus men o die a y ib e in ake did no ma e ially change he esul s o o al Ca in ake, i could be ele an o adjus o hose die a y a iables, aking in o accoun ha consume s wi h a highe consump ion o dai y Fig. 1. Rela i e isk (RR) o di e en diseases o dai y p oduc s in ake. (1) RR o subjec s who consumed $250 g milk daily compa ed wi h pa icipan s who consumed ,70 g/d (Cho e al. 2004). (2) RR o subjec s who con- sumed $25 g co age o ico a cheese daily compa ed wi h hose who did no consume ha kind o cheese (Cho e al. 2004). (3) RR o e e y 100 g liquid dai y p oduc s consumed daily (Missme e al. 2002). (4) Odds o high- es . lowes milk consump ion ca ego ies (Qin e al. 2004). (5, 6) Rela i e odds o highes . lowes milk consump ion ca ego ies (Elwood e al. 2004). Dai y p oduc s and heal h: e iew o e idence S95 h ps://www.camb idge.o g/co e/ e ms. h ps://doi.o g/10.1079/BJN20061709 Downloaded om h ps://www.camb idge.o g/co e. ULPGC. Biblio eca Uni e si a ia, on 24 No 2017 a 09:33:19, subjec o he Camb idge Co e e ms o use, a ailable a p oduc s could ha e o he heal hy die a y habi s, such as a die ich in ui , ege ables and ib e. B eas cance In he p esence o g owing e idence abou he impo ance o he en i onmen in he isk o b eas cance , many s udies analysing he die a y ac o s in his isk ha e been unde aken. Among hese die a y ac o s, dai y p oduc s we e he main a iable analysed in mo e han wen y a icles. Howe e , esul s we e con adic o y. Wi h he pu pose o ge ing b oade e idence, some 10 yea s ago a Canadian eam o esea che s published he i s me a-analysis o combine he esul s o i e coho and wel e case–con ol s udies abou dai y p oduc consump ion and b eas cance isk (Boyd e al. 1993). This me a-analysis ound a small inc eased isk o b eas cance in women wi h g ea e in akes o milk, wi h RR o 1·17 (95 % CI 1·04, 1·30). Since his wo k was published, mo e s udies ha e been unde - aken wi h di e en esul s. Almos 10 yea s la e , in 2002 and o ming pa o he Pooling P ojec , a pooled analysis o eigh coho s udies was published by Missme e al. (2002). Fo his pooled analysis, eigh p ospec i e s udies wi h a leas 200 cases o b eas cance each we e included, b inging a o al o 351041 heal hy women and 7379 in asi e b eas cance cases. The inclusion c i e ia we e simila o hose men ioned o he col- o ec al cance pooled analysis. Wi h espec o dai y p oduc s, di e en ood g oups we e conside ed. Dai y p oduc s we e di ided in o solid (bu e and cheese) o liquid (milk, yoghu , ice c eam, e c.). Subg oups we e also conside ed (whole, semi- skimmed o skimmed milk). Thus, en di e en subg oups o dai y p oduc s we e conside ed. Dai y p oduc in ake was ana- lysed as a con inuous a iable (wi h inc emen s o 100 g daily o all p oduc s excep bu e and c eam, and 10 g o he la e ) and as ca ego ical a iable compa ing highe . lowe qua iles o in ake. O he die a y and non-die a y ac o s associa ed wi h b eas cance we e also conside ed, like o al ene gy o alcohol in ake, pa i y, menopausal s a us and BMI. Simila o he da a men ioned in he colo ec al cance pooling p ojec , he geog aphical a iabili y achie ed he e pe mi ed a wide ange o in a-indi idual a ia ion in dai y p oduc con- sump ion. In ha way, women in he ou h qua ile o liquid dai y p oduc s consumed almos 630 g o his kind o dai y p oduc e e y day, whe eas women in he i s qua ile consumed 360 g. As shown in Fig. 1, esea che s did no ind any ela ionship be ween dai y p oduc in ake and b eas cance isk, nei he ea ing dai y p oduc s as a con inuous a iable no ea ing i as a ca ego ical one. They did no ind s a is ically signi i- can associa ions in any o he en subg oups o dai y p oduc s conside ed, ei he . They concluded ha , aking in o accoun da a o mo e han 350 000 women, he e was no e idence ha a die ich in dai y p oduc s du ing middle o ad anced age could inc ease o modi y he isk o b eas cance in No h Ame ican o Eu opean women. In a simila way, Moo man & Te y (2004) in hei ecen e iew abou consump ion o dai y p oduc s and isk o b eas cance concluded ha published epidemiological da a do no p o ide consis en e idence o an associa ion be ween he consump ion o dai y p oduc s and b eas cance isk. They men ioned some limi a ions ha mus be conside ed. One is he mode a e eliabili y o he me hods used o assess dai y p oduc in ake, which could lead o some misclassi i- ca ion. This misclassi ica ion could make i mo e di icul o de ec he ela i ely small e ec s o mos die a y ac o s on cance isk, which is especially impo an in p ospec i e s udies, whe e he in ake o dai y p oduc s could change g ea ly o e he long ollow-up pe iod. O he limi a ions men ioned by au ho s is ha consump ion o dai y p oduc s may be associa ed wi h o he die a y habi s o o he a iable nu ien con en o dai y p oduc s (such as i amin D, subjec o whe he he ood is o i ied o no ) ha could also in luence b eas cance isk. All hese ac o s should be conside ed in he adjus ed model. P os a e cance P os a e cance is he mos p e alen cance in men in Eu ope and he USA (Quinn & Babb, 2002; Jemal e al. 2005). I is no known which ac o s induce his cance and which agen leads o la en p os a e umou s – whose incidence is e y high in men wo ld- wide – p og essing o a clinically de ec able p os a e cance , he incidence o which, by con as , shows a high a ia ion in di e en geog aphical a eas. The e a e di e en isk ac o s p o- posed ha included ho monal ac o s, amilial an eceden s and die a y a iables. Among die a y a iables, a consump ion was o g ea in e es because animal s udies sugges ed ha a in ake, in combina ion wi h o he die a y a iables, could inc ease umou g ow h. Dai y p oduc in ake was conside ed in h ee di e en e iews all published in 2004. In Canada, Fleshne e al. (2004) e iewed a icles pub- lished o e he las wo decades looking a he ela ionship be ween p os a e cance and a in ake ( o al a in ake o a in speci ic oods like dai y p oduc s). They analysed se en coho s udies, including mo e han 180 000 subjec s, and only one showed a s a is ically signi ican associa ion, once adjus ed; he o he six ound no associa ion. The Heal h P o- essionals Follow-Up S udy obse ed ha subjec s wi h highe in akes o dai y p oduc s had wice he isk o me a- s a ic p os a e cance han subjec s wi h low in ake, bu he au ho s men ioned ha mos o he excess isk could be explained by known nu i ional componen s o hese oods (Ca and a y acids). The Ne he lands coho s udy showed a posi i e end in p os a e cance isk o consump ion o milk p oduc s (P¼0·02), bu no o Ca in ake. A me a-analysis o ele en case–con ol s udies compa ing incidence o p os a e cance in highe . lowe consume s o milk was published by Qin e al. (2004). The combined es ima ed odds a io (OR) was 1·68 (95 % CI 1·34, 2·12), sligh ly smalle when adjus ed o some a iables: 1·56 (95 % CI 1·30, 1·83; Fig. 1). In he discussion, he au ho s men ioned ha eigh s udies used hospi al-based con ols, which a e mo e p one o bias han popula ion-based con ols. Also, milk consump ion was di icul o classi y because o he ambigui y o ood i ems used in some ques ionnai es. They declined o make a me a-analysis o coho s udies because all we e de eloped in Ame ica and no h- e n Eu ope, whe e he high le els o consump ion o dai y p o- duc s would make i di icul o ind a g oup ee om dai y p oduc consump ion. Ano he di icul y men ioned o he inclusion o coho s udies in he me a-analysis was ha in hese s udies he endpoin was no always he same. Finally, in 2004, Dagnelie e al. e iewed p ospec i e coho and in e en ion s udies o analyse he e ec ha E.-E. A ´l a ez-Leo ´ne al.S96 h ps://www.camb idge.o g/co e/ e ms. h ps://doi.o g/10.1079/BJN20061709 Downloaded om h ps://www.camb idge.o g/co e. ULPGC. Biblio eca Uni e si a ia, on 24 No 2017 a 09:33:19, subjec o he Camb idge Co e e ms o use, a ailable a indi idual oods and nu ien s had on p os a e cance . The au ho s excluded case–con ol s udies due o he inconclusi e esul s ob ained and o he g ea e possibili y o selec ion and in o ma ion bias. They iden i ied wen y-se en s udies ha ana- lysed ela ionship be ween milk and dai y p oduc consump ion and p os a e cance isk. Resul s we e equally dis ibu ed be ween a posi i e associa ion ( hi een epo ed) and a null esul ( ou een epo ed). Howe e , only i e o hese p ospec- i e s udies men ioned ha he die a y ques ionnai e used was alida ed. O hese i e s udies, wo had s a is ically signi ican esul s. The Heal h P o essionals Follow-Up S udy showed an inc eased isk o p os a e cance o di e en dai y p oduc s. The RR o highe qua iles o in ake . lowe qua iles anged om 1·06 o co age/ ico a cheese o 1·25 o milk (bo h whole and skimmed o low- a ) and o 1·42 o bu e . A No we- gian s udy o mo e han 25 000 men showed ha men consuming skimmed milk had an RR o 2·2 (95 % CI 1·3, 3·7) o p os a e cance compa ed wi h men consuming whole milk. The o he h ee s udies ound no signi ican associa ion be ween dai y p o- duc in ake and p os a e cance isk. Based on hese esul s, one o he conclusions men ioned by he au ho s is ha he a ailable e idence o speci ic dai y p oduc s emains limi ed o inconclu- si e, which p e en s i m conclusions on ha ood p oduc . One impo an limi a ion men ioned is ha s udies used incidence, mo ali y, o bo h, as an endpoin . In ae iological s udies, inci- dence is mo e app op ia e because mo ali y could be de e - mined by o he non-die a y a iables (such as ea men ). A me a-analysis de eloped wi h coho s udies o e en an in e en ion s udy would cla i y he ela ionship be ween o al a in ake and p os a e cance in o de o es ablish public heal h ecommenda ions. Dai y p oduc s and os eopo osis/bone heal h The e ha e been wo me a-analyses o e iew a icles on bone heal h and dai y p oduc in ake, he mos ecen one by Weinsie and K umdieck published in 2000. They ound wel e a icles wi h he highes le el o e idence, i.e. a an- domized ial wi h a con ol g oup o coho s udies wi h $3000 pa icipan s and an a e age ollow-up pe iod o 5 yea s and esul s adjus ed by mo e han h ee ac o s, such as age, sex, BMI, physical ac i i y, menopausal s a us and/o subs i u i e ho monal he apy. The au ho s did no de elop a me a-analysis o hese s udies, and conclusions we e based on he compa ison o he amoun o s udies wi h posi i e, nega i e o non-de ini i e esul s. Among he wel e a icles, he e we e six ha ound no e ec be ween he in ake o dai y p oduc s and bone heal h, i e ound a posi i e e ec and one ound a nega i e e ec . Women unde 30 yea s seem o gain he mos bene i . Al hough s udies which ana- lysed he composi ion o dai y p oduc s a e sca ce, milk seems o be he mos p o ec i e. The au ho s men ioned limi- a ions o he wo k published up o ha momen , bo h in me h- odology and sample size. These limi a ions make i di icul o ob ain conclusions and i m e idence abou he e ec s o dai y p oduc consump ion on bone heal h. I is impo an o men- ion ha some esul s a e s a is ically signi ican bu clinically o li le ele ance because Ca in ake could be esponsible o only 1 % o he in e -indi idual a iabili y in bone mass. Ano he di icul y is ha wha is impo an o he le el o densi y o he bone sys em is no he p esen dai y in ake, bu he habi ual in ake h oughou li e. In addi ion, as a u he limi a ion men ioned by o he au ho s, a icles mus ake in o accoun he p esence o o he die a y and li e-s yle a iables ha could in luence bone heal h, such as ui in ake, soda be e ages, physical ac i i y, smoking and alcohol in ake. In 1997, Cumming and Ne i published a sys ema ic e iew o assess he e ec i eness o Ca (as supplemen s o om die a y sou ces) o he p e en ion o os eopo o ic ac- u es in pos menopausal women. This wo k was mo e es ic ed han he p e ious one because only s udies wi h ac- u e ou come we e eligible. They ound ou een s udies ha e alua ed Ca supplemen s ( ou andomized ials, h ee non- andomized ials and se en obse a ional epidemiological s udies) and wen y- h ee obse a ional s udies ha e alua ed die a y Ca (non- andomized ials). All ou andomized ials epo ed a educed isk among women ecei ing Ca sup- plemen s. The educ ion in ac u e isk anged om 26 o 70 %, wi h a mean Ca dose o 1050 mg/d. The au ho s men- ioned some limi a ions; in one ial subjec s we e aking Ca and i amin D 3 a he same ime, and in o he s he de ini ion o ac u e di e ed ( adiological educ ion in e eb al body weigh ; a io o an e io o pos e io e eb al body heigh ; o symp oma ic ac u es o w is and hands). The au ho s could use he me a-analysis me hod o analyse esul s o six- een o he obse a ional s udies ha e alua ed he e ec o die a y Ca, al hough among hese s udies indi idual indings we e no consis en because o he he e ogenei y. Pooling he esul s o hese six een s udies, au ho s eached an OR o hip ac u e o 0·96 (95 % CI 0·93, 0·99) pe 300 mg/d inc ease in die a y Ca and 0·88 (95 % CI 0·80, 0·97) pe 1000 mg/d. 300 mg is equi alen o one glass o milk daily and 1000 mg is he amoun in a ypical Ca supplemen . I hey es ic ed analysis o he i e coho s udies, he pooled OR o 300 mg/d was also 0·96, bu he 95 % CI was no longe signi ican ( om 0·91 o 1·02). The main conclusion emphasized by he au ho s was ha inc eased Ca in ake among pos menopausal women appea ed o be associa ed wi h a small educ ion in isk o ac u e, wi h abou a 30 % educ ion in ac u e isk among hose aking highe amoun s o Ca (1 g daily as a supplemen ). The au ho s men ioned ha s a is ical pooling o esul s does no emo e any scien i ic bias in he pooled s udies, and men ioned he need o de elop la ge s udies wi h symp oma ic ac u es as he ou come. Dai y p oduc s and ca dio ascula diseases CVD is a b oad g oup ha emb aces hea diseases (including co ona y diseases and myoca dial in a c ion) and ascula diseases (including hype ension and ce eb o ascula acciden o s oke). The e ha e been almos 100 sys ema ic e iews in ol ing CVD and dai y p oduc s, be ween o he die a y and non-die a y a iables. Howe e , only six we e mo e speci ically o ien ed o he in ake o dai y p oduc s. Dai y p oduc s and hype ension Hype ension is a majo isk ac o o CHD and s oke, wi h a high p e alence in Eu ope. A wide e iew published in a supplemen o he B i ish Jou nal o Nu i ion by He mansen (2000) demons a ed ha he p e en ion and managemen o hype ension – a indi idual and a popula ion le el – should Dai y p oduc s and heal h: e iew o e idence S97 h ps://www.camb idge.o g/co e/ e ms. h ps://doi.o g/10.1079/BJN20061709 Downloaded om h ps://www.camb idge.o g/co e. ULPGC. Biblio eca Uni e si a ia, on 24 No 2017 a 09:33:19, subjec o he Camb idge Co e e ms o use, a ailable a be a p io i y o public heal h au ho i ies. One essen ial s ep o achie e his objec i e by he leas in usi e means is die a y modi ica ion. The e iew ied o explain ha he e idence was ha die a y modi ica ions in luence he p opensi y o de elop hype ension. I was poin ed ou ha he e was no di ec e idence ha educing blood p essu e h ough die a y measu es educes he isk o CVD. In his e iew die a y modi- ica ion was e-examined. Two me a-analyses pooled da a om andomized con olled ials o assess he e ec s o die a y Ca supplemen a ion on blood p essu e (He mansen, 2000). Bo h showed a change in sys olic blood p essu e (SBP) be ween 20·5 and 21·7 mmHg (highe educ ion in hype ensi e sub- jec s), bu dias olic blood p essu e (DBP) was no a ec ed. One poin men ioned in he e iew is ha , wi h he excep ion o alcohol consump ion, he e we e no clinical in e en ion s udies abou he ela ionship be ween mac onu ien in ake and blood p essu e egula ion. They men ioned one inhe en p oblem in s udying he e ec o a change in a speci ic mac onu- ien . In o de o main ain o al ene gy in ake, subjec s mus simul aneously change he in ake o o he mac onu ien s and he genuine cause o he heal h ou come could, he e o e, emain unce ain. Besides changes in one mic o- o mac onu i- en , changes in o e all die a y pa e n may be o g ea alue. The Die a y App oach o S op Hype ension (DASH) s udy was speci ically men ioned in he e iew. The DASH s udy is a suc- cess ul andomized con olled clinical ial ha s a ed in 1995 and has shown ha a die ich in low- a dai y p oduc s, ui s and ege ables could lowe blood p essu e le els. The in e en- ion consis ed o ollowing a ‘combina ion die ’ o 3 weeks. I was ich in ui and ege ables ( en se ings daily) and ich in low- a dai y p oduc s ( h ee se ings daily in ol ing 360 ml skimmed o semi-skimmed milk, 86 g yoghu and 39 g cheese). This combina ion die achie ed a educ ion in blood p essu e in 2 weeks, sugges ing ha complex die a y changes may ha e g ea e e ec s on blood p essu e han single-nu ien modi ica ion. In he conclusions, he au ho men- ioned ha li e-s yle changes p oposed o main ain adequa e blood p essu e con ol migh ha e a wide ange o addi ional bene icial e ec s, which is no ue o mos an ihype ensi e d ugs. In 2003, a e iew ha summa ized he esul s o he DASH die was published by C addick e al. Th ee o ou daily se - ings o dai y p oduc s could diminish SBP in hype ensi e and non-hype ensi e subjec s by 22·7 mmHg (97·5 % CI 24·6, 20·9) and DBP by 21·9 mmHg (97·5 % CI 23·3, 20·6), compa ed wi h a die ich in ui s and ege ables bu low in dai y p oduc s. On his basis, he au ho s ec- ommended ha subjec s wi h blood p essu e o $120/ 80 mmHg ollow heal hy li e-s yle habi s, such as main aining heal hy weigh , educing Na in ake, inc easing egula physi- cal ac i i y, limi ing alcohol in ake and should ollow he DASH die , which is high in ui s, ege ables and low- a dai y p oduc s and educed in a . This die has been shown o educe blood p essu e signi ican ly and also has been shown o educe blood choles e ol and homocys eine le els and o enhance he bene i s o an ihype ensi e d ug he apy. Ano he e iew published by Mille e al. (2000) wen o e he DASH s udy his o y and also made a e ision o he p io e i- dence o obse a ional s udies poin ed ou in he li e a u e abou die a y Ca in ake and hype ension. Cappuccio e al.ina me a-analysis ha was e ised by Bi ke 3 yea s la e , summa ized he obse a ional s udies in 1995. This me a-analysis es ima ed a pooled educ ion o 20·39 mmHg (wi h a ange o 20·47 o 20·31) in SBP, and o 20·21 mmHg ( ange 0·67 o 20·02) in DBP, pe 100 mg die a y Ca inc ease. A he same ime, Buche and co-wo ke s unde ook ano he me a-analysis o andomized con olled ials o Ca supplemen a ion o blood p essu e con ol, and his me a-analysis was also upda ed some yea s la e . Al hough Ca supplemen s a e no in he scope o he p esen e iew, analysis o o y- wo eligible s udies e ealed sig- ni ican educ ions in SBP (21·44 mmHg, 95 % CI 22·2, 0·68) and also in DBP, 20·84 mmHg (95 % CI 21·44, 20·24). Looking a he published e idence, a die low in Na and wi h adequa e in akes o Ca, Mg and K is ecommended o p e en hype ension. Low- a dai y p oduc s ul il his ecommenda ion. Dai y p oduc s and s oke Hype ension is he main isk ac o o s oke. Howe e , nea ly 50 % o s okes occu in subjec s wi h no mal o in he high– no mal limi o blood p essu e. In 2001, a e iew o dai y ood consump ion, blood p essu e and s oke was published by Massey. Two s udies on dai y p oduc s and s oke we e men- ioned. The Honolulu s udy showed ha men who did no con- sume milk su e wice as many s okes han men who consume wo glasses o mo e o milk daily (one glass ¼240 ml). When hey analysed he o igin o he Ca (die a y o as sup- plemen ), hey obse ed ha only die a y Ca was p o ec i e agains s oke. This may be due o he p esence o o he p o ec- i e componen s in dai y p oduc s, o o he unde lying adequa e die a y Ca in ake in hose who also use Ca supplemen s. In he Nu ses’ Heal h S udy, in akes o h ee mine als ha a e p esen s in dai y p oduc s (Ca, Mg and K) we e associa ed wi h lowe isk o ic us, wi h an RR a ound 0·7. This nega i e associa ion was s onge o he Ca ha came om dai y p oduc s han o he non-dai y Ca. Ano he e iew o nu i ional ac o s and s oke was published by Ga iballa (2000). The au ho e alua ed one s udy ha showed a p o ec i e e ec o milk consump ion on he isk o ischaemic s oke in middle-aged men included in he Honolulu Hea P og am. Inc eases in he in ake o die a y Ca om dai y sou ces appea ed o ha e a p o ec i e e ec agains h omboembolic s oke ha was independen o conco- mi an isk ac o s (P,0·05). The e was a wo old excess isk o s oke in men who we e no milk-d inke s . hose who con- sumed nea ly 500 ml/d o mo e (P,0·05). In his s udy, Ca con- sumed om non-dai y sou ces did no show a p o ec i e e ec , unde lying he impo ance o o he a iables ela ed o milk con- sump ion. One limi a ion o his s udy is ha he Honolulu coho showed lowe le els o dai y p oduc in ake han o he popu- la ion-based samples, so he e ec s o g ea e in akes o die a y Ca ha a e commonly obse ed in o he communi ies (o he e ec s o die a y supplemen a ion) could no be examined. Al hough wi h limi a ions, e idence suppo s he hypo hesis ha he in ake o dai y p oduc s could be associa ed wi h a lowe isk o ce eb o ascula acciden o s oke. I is di icul o associa e he p o ec i e e ec o dai y p oduc s o a single mic onu ien . An adequa e me abolic balance be ween Ca, Mg and K is impo an . Dai y p oduc s a e good ood sou ces o hese mine als. Mo e s udies o e alua e he e ec s o he in ake o dai y p oduc s o e long pe iods a e necessa y. In o de o es ablish adequa e ecommenda ions o ic us E.-E. A ´l a ez-Leo ´ne al.S98 h ps://www.camb idge.o g/co e/ e ms. h ps://doi.o g/10.1079/BJN20061709 Downloaded om h ps://www.camb idge.o g/co e. ULPGC. Biblio eca Uni e si a ia, on 24 No 2017 a 09:33:19, subjec o he Camb idge Co e e ms o use, a ailable a p e en ion, p ospec i e s udies mus analyse o e all die a y pa e n and ood consump ion, a he han he in ake o nu ien s. Dai y p oduc s and hea diseases A coho s udy e iew on milk consump ion and IHD was pub- lished ecen ly (Elwood e al. 2004). The au ho s included en p ospec i e coho s udies in which inciden IHD we e ela ed wi h milk in ake. A pooled es ima e o he ela i e odds o IHD in he subjec s wi h he highes milk in akes was 0·87 (95 % CI 0·74, 1·03), compa ed wi h hose wi h he lowes in akes. Au ho s also calcula ed a pooled es ima e o he odds o s oke, ha was 0·83 (95 % CI 0·77, 0·90) o he same subjec s (Fig. 1). Only h ee s udies adjus ed o o al ene gy in ake. The pooled es ima e o he odds o a ascula e en (IHD and/o ischaemic s oke) was 0·85 (95 % CI 0·70, 1·03). The au ho s men ioned ha milk consump ion is no associa ed wi h any inc eased isk o hea disease o s oke. Conclusions The e a e e hical and me hodological limi a ions o he de el- opmen o andomized con olled clinical ials in public heal h nu i ion. In o de o e alua e he e ec o any die a y in e en ion, i is e y impo an o sol e he limi a ions o he a ailable me hodologies. I is possible o de elop mo e s anda - dized me hods o ecall die a y a iables. I has been men ioned ha di e en die a y assessmen me hods migh yield di e en esul s, due o he misclassi ica ion o subjec s. Mo e in es i- ga ions should be unde aken in o de o de elop ools ha a oid his limi a ion in he assessmen o die a y a iables. I is impo an o accumula e he esul s o di e en s udies, in a simila way o he Pooling P ojec (Boyd e al. 1993; Cho e al. 2004). In he Pooling P ojec , he e we e no pa icipan s om Medi e anean and sou h Eu opean coun ies. In hese coun ies, he in ake o dai y p oduc s and o he die a y habi s a e qui e di e en o hose o no h Eu opean and No h Ame ican coun ies. Due o he wide a ia ion in he a e age in ake o dai y p oduc s, he same indi idual could be conside ed a big consume o dai y p oduc s in one con ex and a low consume in ano he . Dai y p oduc s nu ien composi ion could a y depending on he coun y, due o he di e en o i ica ion policy ollowed in each one. P omo ing p ospec i e s udies in di e en coun ies (such as he Eu opean P ospec i e In es i- ga ion in o Cance and Nu i ion and he Supplemen a ion en Vi amines e Mine ´ aux An ioxydan s s udies in sou he n Eu ope) and including he da a in he pooling p ojec , could lead o a be e unde s anding o he ela ionship be ween dai y p oduc s and heal h. Some conside a ions based on he s udies men ioned could imp o e u u e s udies. I is e y impo an o conside all die a y and non-die a y a iables ha could in luence disease isk. The consump ion o dai y p oduc s is likely o be associa ed wi h highe o lowe in akes o o he nu ien s and i will be associa ed wi h heal h beha iou o o he bene icial ac o s which would be di icul o iden i y. I is also impo an o de ine he bes endpoin o e e y disease. Fo example, symp o- ma ic ac u es would be a good endpoin in bone heal h s udies and cance incidence is a be e endpoin han cance mo ali y in cance s udies. The e is ano he inhe en p oblem when he e ec o a change in a speci ic mac onu ien is unde s udy. In o de o main ain he o al ene gy in ake, subjec s mus simul- aneously change he in ake o o he mac onu ien s. Fo his eason, he genuine cause o he heal h ou come could emain unce ain. I is also e y impo an o e alua e he e ec s o he in ake o dai y p oduc s o e long pe iods. In o de o es ablish adequa e ecommenda ions o disease p e- en ion, p ospec i e s udies mus analyse o e all die a y pa - e n and ood consump ion, be e han nu ien in ake. When ecommenda ions on die a y habi s a e made, i should be poin ed ou ha li e-s yle changes p oposed o dec ease isks may ha e an added alue. These li e-s yle changes ha e a wide ange o addi ional bene icial e ec s, which canno be said o be ue o mos d ugs. Re e ences Boyd NF, Ma in LJ, No el M, Lockwood GA & T ichle DL (1993) A me a-analysis o s udies o die a y a and b eas cance isk. B J Cance 68, 627–636. Cho E, Smi h-Wa ne SA, Spiegelman D, e al. (2004) Dai y oods, Ca, and colo ec al cance : a pooled analysis o 10 coho s udies. J Na l Cance Ins 96, 1015–1122, E a um in J Na l Cance Ins 96, 1724. C addick SR, Elme PJ, Oba zanek E, Vollme WM, S e key LP & Swain MC (2003) The DASH die and blood p essu e. Cu A he - oscle Rep 5, 484–491. Cumming RG & Ne i MC (1997) Calcium o p e en ion o os eo- po o ic ac u es in pos menopausal women. J Bone Mine Res 12, 1321–1329. Dagnelie PC, Schuu man AG, Goldbohm RA & Van den B and PA (2004) Die , an h opome ic measu es and p os a e cance isk. BJU In 98, 1939–1950. Elwood PC, Picke ing JE, Huges J, Fehily AM & Ness AR (2004) Milk d inking, ischaemic hea disease and ischaemic s oke II. E idence om coho s udies. Eu J Clin Nu 58, 718–724. Fleshne N, Bagnell PS, Klo z L & Venka eswa an V (2004) Die a y a and p os a e cance . J U ol 171, S19–S24. Ga iballa SE (2000) Nu i ional ac o s in s oke. B J Nu 84, 5–17. He mansen K (2000) Die , blood p essu e and hype ension. B J Nu 83, S113–S119. Jemal A, Mu ay T, Wa d E, Samuels A, Tiwa i RC, Gha oo A, Feue EJ & Thun MJ (2005) Cance s a is ics. CA Cance J Clin 55, 10–30. Massey LK (2001) Dai y ood consump ion, blood p essu e and s oke. J Nu 131, 1875–1878. Mille GD, DiRienzo DD, Reusse ME & McCa on DA (2000) Bene i s o dai y p oduc consump ion on blood p essu e in humans: a sum- ma y o he biomedical li e a u e. J Am Coll Nu 19, S147–S164. Missme SA, Smi h-Wa ne SA, Spiegelman D, e al. (2002) Mea and dai y ood consump ion and b eas cance : a pooled analysis o coho s udies. In J Epidemiol 31, 78–85. Moo man PG & Te y PD (2004) Consump ion o dai y p oduc s and he isk o b eas cance : a e iew o he li e a u e. Am J Clin Nu 80, 5–14. Qin LQ, Xu JY, Wang PY, Kaneko T, Hoshi K & Sa o A (2004) Milk consump ion is a isk ac o o p os a e cance : me a-analysis o case–con ol s udies. Nu Cance 48, 22–27. Quinn M & Babb P (2002) Pa e ns and ends in p os a e cance incidence, su i al, p e alence and mo ali y. Pa I: in e na ional compa isons. BJU In 90, 162–173. Weinsie RL & K umdieck CL (2000) Dai y oods and bone heal h: examina ion e idence. Am J Clin Nu 72, 681–689. Dai y p oduc s and heal h: e iew o e idence S99 h ps://www.camb idge.o g/co e/ e ms. h ps://doi.o g/10.1079/BJN20061709 Downloaded om h ps://www.camb idge.o g/co e. ULPGC. Biblio eca Uni e si a ia, on 24 No 2017 a 09:33:19, subjec o he Camb idge Co e e ms o use, a ailable a