In Defense of Food

Home > Nonfiction > In Defense of Food > Page 7
In Defense of Food Page 7

by Michael Pollan


  What’s go­ing on he­re? We don’t know. It co­uld be the va­ga­ri­es of hu­man di­ges­ti­on. May­be the fi­ber (or so­me ot­her com­po­nent) in a car­rot pro­tects the an­ti­oxi­dant mo­le­cu­le from dest­ruc­ti­on by sto­mach acids early in the di­ges­ti­ve pro­cess. Or it co­uld be we iso­la­ted the wrong an­ti­oxi­dant. Be­ta is just one of a who­le slew of ca­ro­te­nes fo­und in com­mon ve­ge­tab­les; may­be we fo­cu­sed on the wrong one. Or may­be be­ta-ca­ro­te­ne works as an an­ti­oxi­dant only in con­cert with so­me ot­her plant che­mi­cal or pro­cess; un­der ot­her cir­cums­tan­ces it may be­ha­ve as a pro-oxi­dant.

  Inde­ed, to lo­ok at the che­mi­cal com­po­si­ti­on of any com­mon fo­od plant is to re­ali­ze just how much comp­le­xity lurks wit­hin it. He­re’s a list of just the an­ti­oxi­dants that ha­ve be­en iden­ti­fi­ed in a le­af of gar­den-va­ri­ety thyme:

  ala­ni­ne, anet­ho­le es­sen­ti­al oil, api­ge­nin, as­cor­bic acid, be­ta-ca­ro­te­ne, caf­fe­ic acid, camp­he­ne, car­vac­rol, chlo­ro­ge­nic acid, chryso­eri­ol, de­ru­lic acid, eri­odict­yol, euge­nol, 4-ter­pi­nol, gal­lic acid, gam­ma-ter­pi­ne­ne, isich­lo­ro­ge­nic acid, iso­e­uge­nol, isothy­mo­nin, ka­em­fe­rol, la­bi­atic acid, la­uric acid, li­nalyl ace­ta­te, lu­te­olin, met­hi­oni­ne, myrce­ne, myris­tic acid, na­rin­ge­nin, ros­ma­ri­nic acid, se­le­ni­um, tan­nin, thymol, trytop­han, ur­so­lic acid, va­nil­lic acid.

  This is what you in­gest when you eat fo­od fla­vo­red with thyme. So­me of the­se che­mi­cals are bro­ken down by yo­ur di­ges­ti­on, but ot­hers go on to do va­ri­o­us as-yet-unde­ter­mi­ned things to yo­ur body: tur­ning so­me ge­ne’s exp­res­si­on on or off, per­haps, or in­ter­cep­ting a free ra­di­cal be­fo­re it dis­turbs a strand of DNA de­ep in so­me cell. It wo­uld be gre­at to know how this all works, but in the me­an­ti­me we can enj­oy thyme in the know­led­ge that it pro­bably do­esn’t do any harm (sin­ce pe­op­le ha­ve be­en eating it fo­re­ver) and that it might ac­tu­al­ly do so­me go­od (sin­ce pe­op­le ha­ve be­en eating it fo­re­ver), and even if it do­es not­hing at all, we li­ke the way it tas­tes.

  It’s im­por­tant al­so to re­mind our­sel­ves that what re­duc­ti­ve sci­en­ce can ma­na­ge to per­ce­ive well eno­ugh to iso­la­te and study is su­bj­ect to al­most con­ti­nu­al chan­ge, and that we ha­ve a ten­dency to as­su­me that what we can see is the im­por­tant thing to lo­ok at. The vast at­ten­ti­on pa­id to cho­les­te­rol sin­ce the 1950s is lar­gely the re­sult of the fact that for a long ti­me cho­les­te­rol was the only fac­tor lin­ked to he­art di­se­ase that we had the to­ols to me­asu­re. (This is so­me­ti­mes cal­led par­king-lot sci­en­ce, af­ter the le­gen­dary fel­low who lo­ses his keys in a par­king lot and go­es lo­oking for them un­der the stre­et­light-not be­ca­use that’s whe­re he lost them but be­ca­use that’s whe­re it’s easi­est to see.) When we le­ar­ned how to me­asu­re dif­fe­rent types of cho­les­te­rol, and then trigly­ce­ri­des and C-re­ac­ti­ve pro­te­in, tho­se be­ca­me the im­por­tant com­po­nents to study. The­re will no do­ubt be ot­her fac­tors as yet uni­den­ti­fi­ed. It’s an old story: When Pro­ut and Li­ebig na­iled down the mac­ro­nut­ri­ents, sci­en­tists fi­gu­red that they now un­ders­to­od the na­tu­re of fo­od and what the body ne­eded from it. Then when the vi­ta­mins we­re iso­la­ted a few de­ca­des la­ter, sci­en­tists tho­ught, okay, now we re­al­ly un­ders­tand fo­od and what the body ne­eds for its he­alth; and to­day it’s the polyp­he­nols and ca­ro­te­no­ids that se­em to ha­ve comp­le­ted the pic­tu­re. But who knows what el­se is go­ing on de­ep in the so­ul of a car­rot?

  The go­od news is that, to the car­rot eater, it do­esn’t mat­ter. That’s the gre­at thing abo­ut eating fo­ods as com­pa­red with nut­ri­ents: You don’t ne­ed to fat­hom a car­rot’s comp­le­xity in or­der to re­ap its be­ne­fits.

  The mystery of the an­ti­oxi­dants po­ints up the dan­ger in ta­king a nut­ri­ent out of the con­text of fo­od; sci­en­tists ma­ke a se­cond, re­la­ted er­ror when they at­tempt to study the fo­od out of the con­text of the di­et. We eat fo­ods in com­bi­na­ti­ons and in or­ders that can af­fect how they’re me­ta­bo­li­zed. The car­bohyd­ra­tes in a ba­gel will be ab­sor­bed mo­re slowly if the ba­gel is spre­ad with pe­anut but­ter; the fi­ber, fat, and pro­te­in in the pe­anut but­ter cus­hi­on the in­su­lin res­pon­se, the­reby blun­ting the im­pact of the car­bohyd­ra­tes. (This is why eating des­sert at the end of the me­al rat­her than the be­gin­ning is pro­bably a go­od idea.) Drink cof­fee with yo­ur ste­ak, and yo­ur body won’t be ab­le to fully ab­sorb the iron in the me­at. The oli­ve oil with which I eat to­ma­to­es ma­kes the lyco­pe­ne they con­ta­in mo­re ava­ilab­le to my body. So­me of tho­se com­po­unds in the sprig of thyme may af­fect my di­ges­ti­on of the dish I add it to, hel­ping to bre­ak down one com­po­und or sti­mu­la­te pro­duc­ti­on of an enzy­me ne­eded to de­to­xify anot­her. We ha­ve ba­rely be­gun to un­ders­tand the re­la­ti­ons­hips among fo­ods in a cu­isi­ne.

  But we do un­ders­tand so­me of the simp­lest re­la­ti­ons­hips among fo­ods, li­ke the ze­ro-sum re­la­ti­ons­hip: If you eat a lot of one thing, you’re pro­bably not eating a lot of so­met­hing el­se. This fact alo­ne may ha­ve hel­ped le­ad the di­et-he­art re­se­arc­hers ast­ray. Li­ke most of us, they as­su­med that a bad out­co­me li­ke he­art di­se­ase must ha­ve a bad ca­use, li­ke sa­tu­ra­ted fat or cho­les­te­rol, so they fo­cu­sed the­ir in­ves­ti­ga­ti­ve ener­gi­es on how the­se bad nut­ri­ents might ca­use di­se­ase rat­her than on how the ab­sen­ce of so­met­hing el­se, li­ke plant fo­ods or fish, might fi­gu­re in the eti­ology of the di­se­ase. Nut­ri­ti­on sci­en­ce has usu­al­ly put mo­re of its ener­gi­es in­to the idea that the prob­lems it stu­di­es are the re­sult of too much of a bad thing ins­te­ad of too lit­tle of a go­od thing. Is this go­od sci­en­ce or nut­ri­ti­onist pre­j­udi­ce? The epi­de­mi­olo­gist John Pow­les has sug­ges­ted this pre­di­lec­ti­on is lit­tle mo­re than a Pu­ri­tan bi­as: Bad things hap­pen to pe­op­le who eat bad things.

  But what pe­op­le don’t eat may mat­ter as much as what they do. This fact co­uld exp­la­in why po­pu­la­ti­ons that eat di­ets con­ta­ining lots of ani­mal fo­od ge­ne­ral­ly ha­ve hig­her ra­tes of co­ro­nary he­art di­se­ase and can­cer than tho­se that don’t. But nut­ri­ti­onism en­co­ura­ged re­se­arc­hers to lo­ok be­yond the pos­sibly cul­pab­le fo­od it­self-me­at-to the cul­pab­le nut­ri­ent in the me­at, which sci­en­tists ha­ve long as­su­med to be the sa­tu­ra­ted fat. So they are baf­fled in­de­ed when lar­ge di­etary tri­als li­ke the Wo­men’s He­alth Ini­ti­ati­ve and the Nur­ses’ He­alth Study fa­il to find evi­den­ce that re­du­cing fat in­ta­ke sig­ni­fi­cantly re­du­ces the in­ci­den­ce of he­art di­se­ase or can­cer.

  Of co­ur­se thanks to the low-fat-di­et fad (inspi­red by the sa­me re­duc­ti­onist hypot­he­sis abo­ut fat), it is en­ti­rely pos­sib­le to slash yo­ur in­ta­ke of sa­tu­ra­ted fat wit­ho­ut gre­atly re­du­cing yo­ur con­sump­ti­on of ani­mal pro­te­in: Just drink the low-fat milk, buy the low-fat che­ese, and or­der the chic­ken bre­ast or the tur­key ba­con ins­te­ad of the bur­ger. So did the big di­etary tri­als exo­ne­ra­te me­at or just fat? Un­for­tu­na­tely, the fo­cus on nut­ri­ents didn’t tell us much abo­ut fo­ods. Per­haps the culp­rit nut­ri­ent in me­at and da­iry is the ani­mal pro­te­in it­self, as so­me re­se­arc­hers hypot­he­si­ze. (The Cor­nell nut­ri­ti­onist T. Co­lin Camp­bell ar­gu­es as much in his re­cent bo­ok, The Chi­na Study.) Ot­hers think it co­uld be the par­ti­cu­lar kind of iron in red me­at (cal­led he­me iron) or the nit­ro­sa­mi­nes pro­du­ced
when me­at is co­oked. Per­haps it is the ste­ro­id growth hor­mo­nes typi­cal­ly pre­sent in the milk and me­at; the­se hor­mo­nes (which oc­cur na­tu­ral­ly in me­at and milk but are of­ten aug­men­ted in in­dust­ri­al pro­duc­ti­on) are known to pro­mo­te cer­ta­in kinds of can­cer.

  Or, as I men­ti­oned, the prob­lem with a me­at-he­avy di­et might not even be the me­at it­self but the plants that all that me­at has pus­hed off the pla­te. We just don’t know. But eaters wor­ri­ed abo­ut the­ir he­alth ne­edn’t wa­it for sci­en­ce to set­tle this qu­es­ti­on be­fo­re de­ci­ding that it might be wi­se to eat mo­re plants and less me­at. This of co­ur­se is pre­ci­sely what the McGo­vern com­mit­tee was trying to tell us.

  The ze­ro-sum fal­lacy of nut­ri­ti­on sci­en­ce po­ses anot­her obs­tac­le to na­iling down the ef­fect of a sing­le nut­ri­ent. As Gary Ta­ubes po­ints out, it’s dif­fi­cult to de­sign a di­etary tri­al of so­met­hing li­ke sa­tu­ra­ted fat be­ca­use as so­on as you re­mo­ve it from the tri­al di­et, eit­her you ha­ve dra­ma­ti­cal­ly re­du­ced the ca­lo­ri­es in that di­et or you ha­ve rep­la­ced the sa­tu­ra­ted fat with so­met­hing el­se: ot­her fats (but which ones?), or car­bohyd­ra­tes (but what kind?), or pro­te­in. Wha­te­ver you do, you’ve int­ro­du­ced a se­cond va­ri­ab­le in­to the ex­pe­ri­ment, so you will not be ab­le to at­tri­bu­te any ob­ser­ved ef­fect strictly to the ab­sen­ce of sa­tu­ra­ted fat. It co­uld just as easily be due to the re­duc­ti­on in ca­lo­ri­es or the ad­di­ti­on of car­bohyd­ra­tes or pol­yun­sa­tu­ra­ted fats. For every di­et hypot­he­sis you test, you can const­ruct an al­ter­na­ti­ve hypot­he­sis ba­sed on the pre­sen­ce or ab­sen­ce of the subs­ti­tu­te nut­ri­ent. It gets messy.

  And then the­re is the pla­ce­bo ef­fect, which has al­ways be­de­vi­led nut­ri­ti­on re­se­arch. Abo­ut a third of Ame­ri­cans are what re­se­arc­hers call res­pon­ders-pe­op­le who will res­pond to a tre­at­ment or in­ter­ven­ti­on re­gard­less of whet­her they’ve ac­tu­al­ly re­ce­ived it. When tes­ting a drug you can cor­rect for this by using a pla­ce­bo in yo­ur tri­al, but how do you cor­rect for the pla­ce­bo ef­fect in the ca­se of a di­etary tri­al? You can’t: Low-fat fo­ods sel­dom tas­te li­ke the re­al thing, and no per­son is ever go­ing to con­fu­se a me­at entr­йe for a ve­ge­ta­ri­an subs­ti­tu­te.

  Ma­ri­on Nest­le al­so ca­uti­ons aga­inst ta­king the di­et out of the con­text of the li­festy­le, a par­ti­cu­lar ha­zard when com­pa­ring the di­ets of dif­fe­rent po­pu­la­ti­ons. The Me­di­ter­ra­ne­an di­et is wi­dely be­li­eved to be one of the most he­alth­ful tra­di­ti­onal di­ets, yet much of what we know abo­ut it is ba­sed on stu­di­es of pe­op­le li­ving in the 1950s on the is­land of Cre­te-pe­op­le who in many res­pects led li­ves very dif­fe­rent from our own. Yes, they ate lots of oli­ve oil and mo­re fish than me­at. But they al­so did mo­re physi­cal la­bor. As fol­lo­wers of the Gre­ek Ort­ho­dox church, they fas­ted fre­qu­ently. They ate lots of wild gre­ens-we­eds. And, per­haps most sig­ni­fi­cant, they ate far fe­wer to­tal ca­lo­ri­es than we do. Si­mi­larly, much of what we know abo­ut the he­alth be­ne­fits of a ve­ge­ta­ri­an di­et is ba­sed on stu­di­es of Se­venth-Day Ad­ven­tists, who muddy the nut­ri­ti­onal pic­tu­re by abs­ta­ining from al­co­hol and to­bac­co as well as me­at. The­se ext­ra­ne­o­us but una­vo­idab­le fac­tors are cal­led, aptly, con­fo­un­ders.

  One last examp­le: Pe­op­le who ta­ke sup­ple­ments are he­alt­hi­er than the po­pu­la­ti­on at lar­ge, yet the­ir he­alth pro­bably has not­hing what­so­ever to do with the sup­ple­ments they ta­ke-most of which re­cent stu­di­es ha­ve sug­ges­ted are worth­less. Sup­ple­ment ta­kers tend to be bet­ter edu­ca­ted, mo­re aff­lu­ent pe­op­le who, al­most by de­fi­ni­ti­on, ta­ke a gre­ater than usu­al in­te­rest in per­so­nal he­alth-con­fo­un­ders that pro­bably ac­co­unt for the­ir su­pe­ri­or he­alth.

  But if con­fo­un­ding fac­tors of li­festy­le be­de­vil epi­de­mi­olo­gi­cal com­pa­ri­sons of dif­fe­rent po­pu­la­ti­ons, the sup­po­sedly mo­re ri­go­ro­us stu­di­es of lar­ge Ame­ri­can po­pu­la­ti­ons suf­fer from the­ir own ar­gu­ably even mo­re di­sab­ling flaws. In as­cen­ding or­der of sup­po­sed re­li­abi­lity, nut­ri­ti­on re­se­arc­hers ha­ve three ma­in met­hods for stud­ying the im­pact of di­et on he­alth: the ca­se-cont­rol study, the co­hort study, and the in­ter­ven­ti­on tri­al. All three are se­ri­o­usly fla­wed in dif­fe­rent ways.

  In the ca­se-cont­rol study, re­se­arc­hers at­tempt to de­ter­mi­ne the di­et of a su­bj­ect who has be­en di­ag­no­sed with a chro­nic di­se­ase in or­der to un­co­ver its ca­use. One prob­lem is that when pe­op­le get sick they may chan­ge the way they eat, so the di­et they re­port may not be the di­et res­pon­sib­le for the­ir il­lness. Anot­her prob­lem is that the­se pa­ti­ents will typi­cal­ly re­port eating lar­ge amo­unts of wha­te­ver the evil nut­ri­ent of the mo­ment is. The­se pe­op­le re­ad the news­pa­per too; it’s only na­tu­ral to se­arch for the ca­uses of one’s mis­for­tu­ne and, per­haps, to link one’s il­lness to one’s be­ha­vi­or. One of the mo­re per­ni­ci­o­us as­pects of nut­ri­ti­onism is that it en­co­ura­ges us to bla­me our he­alth prob­lems on li­festy­le cho­ices, impl­ying that the in­di­vi­du­al be­ars ul­ti­ma­te res­pon­si­bi­lity for wha­te­ver il­lnes­ses be­fall him. It’s worth ke­eping in mind that a far mo­re po­wer­ful pre­dic­tor of he­art di­se­ase than eit­her di­et or exer­ci­se is so­ci­al class.

  Long-term ob­ser­va­ti­onal stu­di­es of co­hort gro­ups such as the Nur­ses’ He­alth Study rep­re­sent a big step up in re­li­abi­lity from the ca­se-cont­rol study. For one thing, the stu­di­es are pros­pec­ti­ve rat­her than ret­ros­pec­ti­ve: They be­gin trac­king su­bj­ects be­fo­re they be­co­me ill. The Nur­ses’ Study, which has col­lec­ted da­ta on the eating ha­bits and he­alth out­co­mes of mo­re than one hund­red tho­usand wo­men over se­ve­ral de­ca­des (at a cost of mo­re than one hund­red mil­li­on dol­lars), is con­si­de­red the best study of its kind, yet it too has li­mi­ta­ti­ons. One is its re­li­an­ce on fo­od-fre­qu­ency qu­es­ti­on­na­ires (abo­ut which mo­re in a mo­ment). Anot­her is the po­pu­la­ti­on of nur­ses it has cho­sen to study. Cri­tics (no­tably Co­lin Camp­bell) po­int out that the samp­le is re­la­ti­vely uni­form and is even mo­re car­ni­vo­ro­us than the U.S. po­pu­la­ti­on as a who­le. Pretty much ever­yo­ne in the gro­up eats a Wes­tern di­et. This me­ans that when re­se­arc­hers di­vi­de the su­bj­ect po­pu­la­ti­on in­to gro­ups (typi­cal­ly fifths) to study the im­pact of, say, a low-fat di­et, the qu­in­ti­le eating the lo­west-fat di­et is not all that low-or so dra­ma­ti­cal­ly dif­fe­rent from the qu­in­ti­le con­su­ming the hig­hest-fat di­et. “Vir­tu­al­ly this en­ti­re co­hort of nur­ses is con­su­ming a high-risk di­et,” ac­cor­ding to Camp­bell. That might exp­la­in why the Nur­ses’ Study has fa­iled to de­tect sig­ni­fi­cant be­ne­fits for many of the di­etary in­ter­ven­ti­ons it’s lo­oked at. In a su­bj­ect po­pu­la­ti­on that is eating a fa­irly stan­dard Wes­tern di­et, as this one is, you’re ne­ver go­ing to cap­tu­re the ef­fects, go­od or bad, of mo­re ra­di­cal­ly dif­fe­rent ways of eating. (In his bo­ok, Camp­bell re­ports Wal­ter Wil­lett’s per­so­nal res­pon­se to this cri­ti­cism: “You may be right, Co­lin, but pe­op­le don’t want to go the­re.”)

  The so-cal­led gold stan­dard in nut­ri­ti­on re­se­arch is the lar­ge-sca­le in­ter­ven­ti­on study. In the­se stu­di­es, of which the Wo­men’s He­alth Ini­ti­ati­ve is the big­gest and best known, a lar­ge po­pu­la­ti­on is di­vi­ded in­to two gro­ups. The in­ter­ven­ti­on gro­up chan­ges its di
­et in so­me presc­ri­bed way whi­le the cont­rol gro­up (one ho­pes) do­es not. The two gro­ups are then trac­ked over many ye­ars to le­arn whet­her the in­ter­ven­ti­on af­fects re­la­ti­ve ra­tes of chro­nic di­se­ase. In the ca­se of the Wo­men’s He­alth Ini­ti­ati­ve study of di­etary fat, a $415 mil­li­on un­der­ta­king spon­so­red by the Na­ti­onal Ins­ti­tu­tes of He­alth, the eating ha­bits and he­alth out­co­mes of ne­arly forty-ni­ne tho­usand wo­men (aged fifty to se­venty-ni­ne) we­re trac­ked for eight ye­ars to as­sess the im­pact of a low-fat di­et on a wo­man’s risk of bre­ast and co­lo­rec­tal can­cer and car­di­ovas­cu­lar di­se­ase. Forty per­cent of the wo­men we­re told to re­du­ce the­ir con­sump­ti­on of fat to 20 per­cent of to­tal ca­lo­ri­es. When the re­sults we­re an­no­un­ced in 2006, it ma­de front-pa­ge news (The New York Ti­mes he­ad­li­ne sa­id LOW-FAT DI­ET DO­ES NOT CUT HE­ALTH RISKS, STUDY FINDS) and the clo­ud of nut­ri­ti­onal con­fu­si­on be­ne­ath which Ame­ri­cans en­de­avor to eat dar­ke­ned furt­her.

 

‹ Prev