Integrative development of a short screening questionnaire of highly processed food consumption (sQ-HPF)

Ma inez‑Pe eze al. In J Beha Nu Phys Ac (2022) 19:6 h ps://doi.o g/10.1186/s12966‑021‑01240‑6 RESEARCH In eg a i e de elopmen o asho sc eening ques ionnai e o highly p ocessed ood consump ion (sQ‑HPF) Celia Ma inez‑Pe ez1, Lidia Daimiel1* , C is ina Climen ‑Maina 1, Miguel Ángel Ma ínez‑González2,3,4, Jo di Salas‑Sal adó2,5,6, Dolo es Co ella2,7, Helmu Sch öde 8, Jose Al edo Ma inez2,9,10, Ángel M. Alonso‑Gómez2,11, Julia Wä nbe g2,12, Jesús Vioque13,14, Do a Romague a2,15, José López‑Mi anda2,16, Ramón Es uch2,17, F ancisco J. Tinahones2,18, José Lape a2,19, Lluis Se a‑Majem2,20, Au o a Bueno‑Ca anillas13,21, Josep A. Tu 2,22, Vicen e Ma ín Sánchez13,23, Xa ie Pin ó2,24, Miguel Delgado‑Rod íguez13,25, Pila Ma ía‑Ma ín26, Josep Vidal27,28, Clo ilde Vázquez2,29, Emilio Ros2,28, Ja ie Bas e a3,30, Nancy Babio2,5,6, Pa icia Guillem‑Saiz2,7, Ma ía Dolo es Zomeño8,31, I zia Abe e2,10, Jessica Vaque o‑Luna2,11, F ancisco Ja ie Ba ón‑López32, Sand a Gonzalez‑Palacios13,14, Jadwiga Konieczna2,15, An onio Ga cia‑Rios2,16, Ma ía Rosa Be nal‑López33, José Manuel San os‑Lozano2,19, Mai a Bes‑Ras ollo2,3, Nadine Khou y2,5,6, Ca men Saiz2,7, Ka la Alejand a Pé ez‑Vega8, Ma ía Angeles Zule 2,10, Lucas Tojal‑Sie a2,11, Zenaida Vázquez Ruiz2,3, Ma ia Angeles Ma inez2,5,6, Mi eia Malcampo8, José M. O do ás1,34 and Rod igo San‑C is obal9 Abs ac Backg ound: Recen li es yle changes include inc eased consump ion o highly p ocessed oods (HPF), which has been associa ed wi h an inc eased isk o non‑communicable diseases (NCDs). Howe e , nu i ional in o ma ion elies on he es ima ion o HPF consump ion om ood‑ equency ques ionnai es (FFQ) ha a e no explici ly de eloped o his pu pose. We aimed o de elop a sho sc eening ques ionnai e o HPF consump ion (sQ‑HPF) ha in eg a es c i e ia om he exis ing ood classi ica ion sys ems. Me hods: Da a om 4400 pa icipan s (48.1% emale and 51.9% male, 64.9 ± 4.9 yea s) o he Spanish PREDIMED‑Plus (“PRE en ion wi h MEDi e anean DIe ”) ial we e used o his analysis. I ems om he FFQ we e classi ied acco ding o ou main ood p ocessing‑based classi ica ion sys ems (NOVA, IARC, IFIC and UNC). Pa icipan s we e classi ied in o e iles o HPF consump ion acco ding o each sys em. Using binomial logis ic eg ession, ood g oups associa ed wi h ag eemen in he highes e ile o a leas wo classi ica ion sys ems we e chosen as i ems o he ques ionnai e. ROC analysis was used o de e mine cu ‑o poin s o he equency o consump ion o each i em, om which a sco e was calcula ed. In e nal consis ency o he ques ionnai e was assessed h ough explo a o y ac o analysis (EFA) and C onbach’s analysis, and ag eemen wi h he ou classi ica ions was assessed wi h weigh ed kappa coe icien s. © The Au ho (s) 2022. Open Access This a icle is licensed unde a C ea i e Commons A ibu ion 4.0 In e na ional License, which pe mi s use, sha ing, adap a ion, dis ibu ion and ep oduc ion in any medium o o ma , as long as you gi e app op ia e c edi o he o iginal au ho (s) and he sou ce, p o ide a link o he C ea i e Commons licence, and indica e i changes we e made. The images o o he hi d pa y ma e ial in his a icle a e included in he a icle’s C ea i e Commons licence, unless indica ed o he wise in a c edi line o he ma e ial. I ma e ial is no included in he a icle’s C ea i e Commons licence and you in ended use is no pe mi ed by s a u o y egula ion o exceeds he pe mi ed use, you will need o ob ain pe mission di ec ly om he copy igh holde . To iew a copy o his licence, isi h p:// c ea i eco mmons. o g/ licen ses/ by/4. 0/. The C ea i e Commons Public Domain Dedica ion wai e (h p:// c ea i eco mmons. o g/ publi cdoma in/ ze o/1. 0/) applies o he da a made a ailable in his a icle, unless o he wise s a ed in a c edi line o he da a. Open Access *Co espondence: [email p o ec ed] 1 Nu i ional Genomics and Epigenomics G oup, P ecision Nu i ion and Obesi y P og am, IMDEA Food, CEI UAM + CSIC, C a. Can oblanco, 8, 28049 Mad id, Spain Full lis o au ho in o ma ion is a ailable a he end o he a icle Page 2 o 16 Ma inez‑Pe eze al. In J Beha Nu Phys Ac (2022) 19:6 Backg ound Changes in ea ing pa e ns a e occu ing wo ldwide [1]. A common ea u e o such changes is he ansi ion om minimally p ocessed ood o mode a ely p ocessed o highly p ocessed (HPF) o ul a-p ocessed oods (UPF) [2–10]. A widely used, al hough con o e sial [11], de i- ni ion o UPF is ha hese a e “indus ial o mula ions made mos ly o en i ely om subs ances de i ed om oods and addi i es, wi h li le i any in ac ood” [12]. Acco ding o he NOVA classi ica ion sys em, hese oods a e highly pala able and habi - o ming, mic obio- logically sa e, a o dable, s ongly ma ke ed and ad e - ised, and sold in con enien and a ac i e packaging, p omo ing hei o e consump ion [12, 13]. This, oge he wi h he e idence showing hei nega i e impac on heal h [9, 10, 14–18], has u ned UPF consump ion in o a po en ial public heal h conce n [19] in need o mo e solid esea ch. While he e m UPF is mainly a ibu ed o he NOVA sys em, o he ood p ocessing-based classi ica ion sys ems ha e desc ibed oods and d inks o simila cha - ac e is ics unde hei ca ego ies o HPF [20, 21], so we will use he e m HPF o e e o his ype o oods. Al hough nume ous s udies ha e demons a ed a link be ween he isk o de eloping non-communicable dis- eases and HPF consump ion [12, 17, 22–27], epidemio- logical esea ch aces some hu dles in his ield. Fi s , he exis ence o mul iple ood p ocessing-based classi ica- ion sys ems based on di e en c i e ia [21, 28] esul s in he e ogeneous conclusions ega ding heal h ou comes depending on he sys em used, as shown ecen ly o ca diome abolic heal h ma ke s [20]. Second, he lack o an e ec i e ool o assess HPF consump ion in clini- cal s udies se e ely hinde s he downs eam analysis o i s ela ionship o disease isk. Fo ins ance, in he PREDIMED-Plus ( om he Spanish “PRE en ion wi h MEDi e anean DIe ”) ial, die a y in ake was assessed h ough a leng hy ood equency ques ionnai e (FFQ). The e o e, es ima ion o HPF consump ion equi es he classi ica ion o FFQ i ems acco ding o he selec ed classi ica ion sys em, as p e iously done [20, 24, 25, 29]. This p ocess is ime-consuming and subjec o bias on he pa o he esea che classi ying he i ems since FFQs a e no speci ically designed o include HPF. This is because an FFQ is designed o es ima e consump ion o gene al commonly consumed oods, o which some may all in o he HPF ca ego y, bu his depends on he classi ica ion sys em used [20]. In addi ion, calcula ions o daily consump ion o each HPF i em and pe cen age o e o al in ake (in g ams o kcal pe day) a e commonly used o in e hei associa ion wi h heal h ou comes [9, 17, 30]. These calcula ions a e no di ec and ime-con- suming when de i ed om cu en ools such as FFQs o 24 h ecalls. The e is, he e o e, he need o an easy- o- use and comp ehensi e measu e ha assesses HPF con- sump ion in he gene al popula ion. This pape desc ibes he de elopmen o a new sc eening ques ionnai e ha allows an easy and quick de e mina ion o a subjec ’s HPF consump ion, he sQ-HPF. We iew his as an in e- g a i e ool since i inco po a es c i e ia om ou ood- p ocessing-based classi ica ion sys ems. We de eloped his ques ionnai e based on a ailable da a om an FFQ o c ea e a ool ha could eplace he es ima ion o HPF consump ion om FFQ in u u e s udies. We hypo he- size ha he sQ-HPF is compa able in e ms o e alua ing HPF consump ion o o he die a y assessmen ools ha mus be used in combina ion wi h ood p ocessing-based classi ica ion sys ems and can e ec i ely cap u e longi u- dinal changes in HPF consump ion. This ques ionnai e will be o po en ial in e es o he scien i ic communi y, especially in he con ex o clini- cal nu i ion and public heal h. I s use should minimize he di icul y in compa ing esul s om s udies ha use di e en classi ica ion sys ems, allowing a s aigh - o wa d e alua ion o he HPF die a y pa e n in la ge Resul s: Reg ession analysis iden i ied 14 ood g oups (i ems) associa ed wi h high HPF consump ion o a leas wo classi ica ion sys ems. EFA showed ha i ems we e ep esen a i e con ibu o s o a single unde lying ac o , he “HPF die a y pa e n” ( ac o loadings a ound 0.2). We cons uc ed a ques ionnai e asking abou he equency o consump‑ ion o hose i ems. The h eshold equency o consump ion was selec ed using ROC analysis. Compa ison o he ou classi ica ion sys ems and he sQ‑HPF showed a ai o high ag eemen . Signi ican changes in li es yle cha ac e is ics we e de ec ed ac oss e iles o he sQ‑HPF sco e. Longi udinal changes in HPF consump ion we e also de ec ed by he sQ‑HPF, conco dan ly wi h exis ing classi ica ion sys ems. Conclusions: We de eloped a p ac ical ool o measu e HPF consump ion, he sQ‑HPF. This may be a aluable ins u‑ men o s udy i s ela ionship wi h NCDs. T ial egis a ion: Re ospec i ely egis e ed a he In e na ional S anda d Randomized Con olled T ial Regis y (ISRCT N8989 8870) on July 24, 2014. Keywo ds: Ul a‑p ocessed ood, Highly p ocessed ood, Ques ionnai e, PREDIMED‑Plus, NOVA, Food p ocessing‑ based classi ica ion Page 3 o 16 Ma inez‑Pe eze al. In J Beha Nu Phys Ac (2022) 19:6 epidemiological s udies e icien ly and compa ably. Al oge he , i will enable he de elopmen o ailo ed nu i ional in e en ions and ood policies o limi HPF consump ion and p e en die - ela ed diseases. Me hods S udy popula ion Da a om he PREDIMED-Plus ial ( om he Spanish “PRE en ion wi h MEDi e anean DIe ”) was used. This is an ongoing 6-yea , mul icen e , andomized, pa allel- g oup clinical ial launched in Spain in 2013. The main aim o he s udy is o e alua e he e ec on p ima y ca - dio ascula disease p e en ion o an in ensi e weigh loss and i s long- e m main enance h ough a li es yle in e - en ion based on h ee pilla s: ene gy- es ic ed Medi- e anean die (e -MedDie ), inc eased physical ac i i y (PA) and beha io al suppo . The s udy p o ocol, includ- ing s udy design and da a collec ion, can be ound a he PREDIMED-Plus websi e (h ps:// www. p edi medpl us. com/ en/) and was app o ed acco ding o he e hical s anda ds o he Decla a ion o Helsinki by he Ins i u- ional Re iew Boa ds (IRBs) o all pa icipa ing cen e s. All pa icipan s p o ided w i en consen o hei pa - icipa ion in he s udy. The ial is conduc ed in 23 Span- ish cen e s and in ol es 6874 pa icipan s (48.5% emale, 54.5% male) be ween 55 and 75 yea s old (mean age and SD 65.0 ± 4.9) who p esen ed wi h o e weigh o obesi y (BMI ≥ 27 and < 40 kg/m2) and me a leas h ee c i e ia o me abolic synd ome (Me S) as p e iously desc ibed [31]. De ails abou he coho ha e been desc ibed else- whe e [32]. The ial was e ospec i ely egis e ed a he In e na ional S anda d Randomized Con olled T ial Regis y wi h numbe 89898870 on 24 h July 2014. The p esen analysis used baseline and longi udinal (yea s 1 and 2) da a om he PREDIMED-Plus s udy da a se da ed 26 h June 2020 was used. Pa icipan s wi h implau- sible ene gy in akes (< 500 o > 3500 kcal o emales and < 800 o > 4000 kcal o males) we e excluded. In addi- ion, pa icipan s wi h missing alues o die a y, li e- s yle and socioeconomic a iables we e no included in he analysis. A e he de ini ion o he e ile ag eemen a iable (see S a is ical analyses sec ion), pa icipan s wi h (1) no coincidence in ex eme e iles o HPF con- sump ion (ei he in e ile 1 o e ile 3) acco ding o a leas wo classi ica ion sys ems and (2) pa icipan s ha we e classi ied in e ile 1 o wo classi ica ion sys ems and e ile 3 o he o he wo classi ica ion sys ems we e excluded. This was done o selec ex eme HPF consum- e s o u he cons uc he binomial eg ession model. The inal numbe o pa icipan s included in he analy- sis o he de elopmen o he ques ionnai e was 4400 (48.1% emale and 51.9% male, 64.9 ± 4.9 yea s) (Fig.1). This s udy adhe ed o he STROBE-nu epo ing guide- lines [33]. Blood measu emen s T ained nu ses collec ed blood samples a e o e nigh as ing a he ec ui ing cen e s o p ima y heal h ca e cen e s. Plasma glucose, iglyce ides and o al choles- e ol le els we e measu ed ollowing s anda d enzyma ic me hods. An h opome ic measu emen s Weigh and wais ci cum e ence measu emen s we e aken om pa icipan s in ligh clo hing wi h no shoes o accesso ies, using an elec onic calib a ed scale and an an h opome ic ape, espec i ely. Wais ci cum e ence was measu ed midway be ween he lowes ib and he iliac c es . Heigh measu emen s we e aken using a wall- moun ed s adiome e . Body Mass Index (BMI) was calcu- la ed as he weigh in kilog ams di ided by he squa e o heigh in me e s. Li es yle measu emen s Socioeconomic and PA da a we e collec ed h ough he gene al PREDIMED-Plus ques ionnai e. Based on he alida ed ques ionnai es, he REGICOR [34] and he Rapid Assessmen o Physical Ac i i y (RAPA) [35], pa icipan s we e asked abou he equency and in en- si y o physical ac i i ies, and h ee le els o PA we e de ined as ollows: low ( equen si ing and li le walking and/o equen si ing and mode a e sus ained e o s), medium ( equen walking wi h no igo ous e o s), high ( equen walking and igo ous e o s and/o equen igo ous e o s). Seden a y beha io s we e assessed h ough a alida ed ques ionnai e, he Spanish e sion o he Nu ses’ Heal h S udy (NHS) ques ionnai e [36]. Da a abou ea ing habi s (binge ea ing, snacking) we e col- lec ed wi hin he mul idimensional scale o weigh locus con ol ques ionnai e [37]. T ained in e iewe s admin- is e ed ques ionnai es in indi idual ace- o- ace sessions. Die a y measu emen s To assess he die a y in ake o pa icipan s o e he las yea , a alida ed semi-quan i a i e 143-i em FFQ [38– 40] ha conside s a ia ions in die a y pa e ns among seasons, weekdays and weekends was used. Pa icipan s we e asked he a e age equency o consump ion o a commonly used po ion size (e.g., glass, cup, slice) o each ood o be e age i em. Nine op ions o e- quency o consump ion a e gi en, anging om “ne e o ha dly e e ” o “mo e han six imes a day.” To es i- ma e he daily consump ion o each i em, he po ion size was mul iplied by he equency o consump ion and hen exp essed as g ams pe day. This calcula ion Page 4 o 16 Ma inez‑Pe eze al. In J Beha Nu Phys Ac (2022) 19:6 was no possible o he ied oods i em as he po ion size is no speci ied in he FFQ. To assess adhe ence o an e -MedDie , a 17-i em ques ionnai e specially de eloped and alida ed o he PREDIMED-Plus ial [41, 42] was used. Th ough his ques ionnai e, he e- quency o consump ion o adi ional Medi e anean ood i ems is e alua ed. One poin is sco ed when he answe mee s speci ic c i e ia de ining e -MedDie , so he highe he sco e, he be e adhe ence o his die . HPF consump ion and ood p ocessing‑based classi ica ion sys ems Fou ood-p ocessing based classi ica ion sys ems – he NOVA [12, 43, 44], he In e na ional Agency o Resea ch on Cance (IARC) [7, 45], he In e na ional Food In o ma ion Council Founda ion (IFIC) [46, 47], and he Uni e si y o No h Ca olina (UNC) [48] sys- ems – we e used o classi y FFQ i ems in o p ocessing ca ego ies as p e iously desc ibed [20]. In he p esen Fig. 1 Flowcha o he PREDIMED‑Plus pa icipan s. Numbe o subjec s shown in bold Page 5 o 16 Ma inez‑Pe eze al. In J Beha Nu Phys Ac (2022) 19:6 s udy, HPF e e s o he ollowing g oups: G oup 4 o NOVA, G oup 3 o IARC, G oups 4 and 5 o IFIC, and G oups 4.1 and 4.2 o UNC. Fo each pa icipan , HPF consump ion was es ima ed acco ding o each classi ica- ion sys em as he sum o g ams pe day consumed om oods in he HPF g oup, di ided by he o al g ams o ood consumed pe day and mul iplied by 100 [20]. The equency o consump ion was di ec ly ob ained om FFQ answe s in imes pe day. S a is ical analyses Da a analysis was conduc ed using R p og amming lan- guage [49] in RS udio [50] and wi h he ollowing s a is- ical packages: “DescTools” [51], “psych” [52], “ ableone” [53], “cu poin ” [54], “co plo ” [55], “ cd” [56] and s a isx [57]. Pa icipan s we e classi ied acco ding o e iles o HPF consump ion o each classi ica ion sys- em (T1 – low HPF consump ion, T3 high HPF con- sump ion). Te iles we e chosen o cap u e and show he a iabili y in HPF consump ion while allowing a s aigh - o wa d compa ison be ween classi ica ion sys ems. Then, subjec s we e classi ied acco ding o e ile ag ee- men o , a leas , wo classi ica ion sys ems. Those clas- si ied in T3 in a leas wo classi ica ion sys ems sco ed “1” o e ile ag eemen , while hose classi ied in T1 o a leas wo classi ica ion sys ems sco ed “0”. FFQ i ems we e classi ied in o ood g oups acco ding o hei simi- la na u e, nu i ional cha ac e is ics, and/o o m o consump ion (Supplemen a y Table1). The associa ion be ween e ile ag eemen and equency o consump- ion o each ood g oup was analyzed h ough binomial logis ic eg ession, adjus ed o age, sex, ec ui men cen e , ene gy in ake, physical ac i i y le el, medica ion o blood p essu e and diabe es, wo king s a us, educa- ional le el, and ci il s a us, using he “glm” unc ion om R base package “s a s.” These co a ia es we e selec ed due o hei po en ial e ec on HPF consump ion. To calcu- la e op imal cu -o poin s o equency o consump ion o he selec ed i ems, ecei e ope a ing cha ac e is ic (ROC) analysis was pe o med using he R package “cu - poin ” and he “cu poin ” unc ion [54]. The me hod o es ima e he cu -o poin s was based on he maximiza- ion o he Youden-Index [58]. Es ima ed cu -o poin s we e used o es ablish he c i e ia o sco ing 1 poin in he sQ-HPF, which indica ed high HPF consump ion, o 0, indica ing low HPF consump ion. Cu -o poin s we e adap ed o he nine possible answe s o he FFQ, so he c i e ia o sco ing we e based on a h eshold equency o consump ion o each i em (i.e., ood g oup). In pa - allel, explo a o y ac o analysis (EFA) was pe o med o iden i y unde lying ela ionship pa e ns be ween i ems included in he sQ-HPF using he “ a” unc ion om he “psych” package [52]. To es o da a sui abili y o he EFA, he Kaise -Meye -Olkin C i e ion [59] and Ba le ’s es o sphe ici y [60] we e applied. The EFA was pe - o med wi hou o a ion and wi h a p incipal ac o solu- ion as a ac o ing me hod. Fac o e en ion was based on he sc ee plo and Kaise ’s c i e ion [61]. C onbach’s alpha [62] was calcula ed as a measu e o in e nal con- sis ency o he ques ionnai e. Using he c i e ia o sco - ing, he sQ-HPF sco e was calcula ed o each subjec in he PREDIMED-Plus da abase, and he ques ionnai e es ima ed HPF consump ion was calcula ed h ough linea eg ession analysis using he sQ-HPF sco e as he dependen a iable. Fo desc ip i e analyses, pa icipan s we e classi ied in o e iles o he sQ-HPF sco e. Da a is shown in ables as “mean (s anda d de ia ion, SD)” o con inuous a iables and as “numbe o subjec s (%)” o ca ego ical a iables. S a is ically signi ican di e ences (p< 0.05) in die e ic and li es yle a iables among e iles we e compa ed using a one-way ANOVA es o con in- uous a iables and a Chi-squa ed es o ca ego ical a - iables. P- alues we e adjus ed o age, sex, ec ui men cen e , ene gy in ake, physical ac i i y le el, medica ion o blood p essu e and diabe es, wo king s a us, educa- ional le el, and ci il s a us. To assess he conco dance be ween e iles o HPF consump ion calcula ed by he ou classi ica ion sys ems and by he sQ-HPF, weigh ed Cohen’s kappa (κ) coe icien s we e calcula ed wi h he unc ion “Kappa” om he R package “ cd”. Fo longi- udinal analysis o HPF consump ion, a linea mixed model was pe o med using he R package “lme4” and “emmeans” wi h he same co a ia es as p e ious analy- ses. The numbe o subjec s used o his analysis was 3284 due o longi udinal da a loss. Resul s Gene al cha ac e is ics o  hePREDIMED‑Plus coho acco ding o e ile ag eemen Subjec s sco ed 1 in he e ile ag eemen a iable i hey we e classi ied in he highes e ile (T3) o HPF con- sump ion o a leas wo classi ica ions sys ems, while hey sco ed 0 i hey we e classi ied in he lowes e ile (T1) o HPF consump ion o a leas wo classi ica ions sys ems. Gene al cha ac e is ics o PREDIMED-Plus pa - icipan s a baseline acco ding o he sco es o he e ile ag eemen a iable a e shown in Table1. Subjec s who sco ed 1 (high HPF consump ion by a leas wo classi- ica ion sys ems) we e mainly men (73.1%), had highe ene gy in ake (2559.84 kcal/day) and lowe MedDie adhe ence (7.65 poin s) han subjec s who sco ed 0 (low HPF consump ion by a leas wo classi ica ions sys ems). Among he high HPF subjec s, 78.8% we e ma ied and 41.8% had a p ima y educa ion le el. A ound hal o he subjec s who sco ed 1 showed a low le el o PA (55.3%) and we e aking medica ion o choles e ol (49.8%). Page 6 o 16 Ma inez‑Pe eze al. In J Beha Nu Phys Ac (2022) 19:6 In addi ion, 75.7% o hem we e aking medica ion o blood p essu e and 25.9% we e on diabe es medica ion. Acco ding o hese esul s, he a iables age, sex, ec ui - men cen e , ene gy in ake, physical ac i i y le el, medi- ca ion o blood p essu e and diabe es, wo king s a us, educa ional le el, and ci il s a us we e selec ed as co a i- a es o u he analysis due o hei po en ial e ec on HPF consump ion. MedDie adhe ence was no selec ed as a co a ia e due o he p esence o collinea i y wi h HPF consump ion. De elopmen o  hesQ‑HPF Food g oups we e de ined based on he simila i y in na u e, nu i ional p o ile and/o o m o consump ion o he PREDIMED-Plus FFQ ood and be e age i ems (Sup- plemen a y Table1). Food g oups chosen o be included in he sQ-HPF showed a signi ican posi i e associa ion (p- alue adjus ed by Bon e oni < 0.05) be ween hei e- quency o consump ion and a alue o 1 o e ile ag ee- men , i.e., he subjec is in e ile 3 o HPF consump ion o a leas wo ood p ocessing-based classi ica ion sys- ems (Table2). A inal solu ion o 14 ood g oups was selec ed and included he ollowing: a y dai y p oduc s, suga y dai y p oduc s, cu ed mea , a s, e men ed alco- hol, dis illed alcohol, suga y and a i icially swee ened d inks, swee s, snacks, eady o ea p oduc s, e ined ce eals, sauces, addi i es, and ied oods. EFA e ealed ha mos o he i ems selec ed o he sQ-HPF had ac o loadings highe han 0.2, indica ing ha hey we e ep esen a i e con ibu o s o he ac- o (Table3). The measu e o sample adequacy (MSA) was 0.78, conside ed as “good” o he e i ica ion o he p opo ion o a iance in a iables ha can be caused by ac o s, acco ding o he Kaise -Meye -Olkin C i e- ion. Ba le ’s es o sphe ici y was highly signi ican (p< 0.001), indica ing ha a iables we e co ela ed in he popula ion. This, oge he wi h he MSA alue, indi- ca ed he adequacy o he da a o p oceed wi h EFA. Fac o e en ion applying he Kaise c i e ion e ealed a single unde lying ac o being iden i ied by he ques ion- nai e i ems, namely he HPF die a y pa e n. The in e - nal consis ency o he ques ionnai e i ems was e alua ed wi h C onbach’s alpha, which had a mode a e alue o 0.67. Table 1 Gene al cha ac e is ics o PREDIMED‑Plus pa icipan s a baseline acco ding o e ile ag eemen Da a shown as “mean (s anda d de ia ion, SD)” o con inuous a iables and as “numbe o subjec s (%)” o ca ego ical a iables. One‑way ANOVA es used o con inuous a iables and Chi‑squa ed es used o ca ego ical a iables. Signi ican p‑ alues (< 0.05) shown in bold MedDie Medi e anean die , PA Physical ac i i y a Te ile ag eemen a iable: sco e 0 – “low HPF consume ” i a subjec is classi ied in T1 o HPF consump ion by a leas wo classi ica ion sys ems; sco e 1 – “high HPF consume ” i a subjec is classi ied in T3 by a leas wo classi ica ion sys ems. HPF: highly p ocessed ood. N= 4400 Te ile ag eemen a Low HPF High HPF p n2186 2214 Age (yea s) 66.14 (4.50) 63.68 (5.07) < 0.001 Female sex (%) 1522 (69.6) 595 (26.9) < 0.001 Ci il s a us (%) 0.001 Single 115 (5.3) 109 (4.9) Ma ied 1622 (74.2) 1745 (78.8) Widowed/di o ced 449 (20.5) 360 (16.3) Educa ion le el (%) P ima y 1252 (57.3) 926 (41.8) < 0.001 Seconda y 538 (24.6) 704 (31.8) College 396 (18.1) 584 (26.4) Ac i e wo king s a us (%) 307 (14.0) 642 (29.0) < 0.001 Ene gy in ake (kcal/day) 2158.46 (480.16) 2559.84 (571.11) < 0.001 MedDie adhe ence sco e 9.46 (2.54) 7.65 (2.61) < 0.001 PA le el (%) 0.001 Low 1188 (54.3) 1225 (55.3) Medium 904 (41.4) 841 (38.0) High 94 (4.3) 148 (6.7) Blood p essu e medica ion (%) 1717 (78.5) 1676 (75.7) 0.027 Choles e ol medica ion (%) 1123 (51.4) 1102 (49.8) 0.303 Diabe es medica ion (%) 631 (28.9) 573 (25.9) 0.029 Page 7 o 16 Ma inez‑Pe eze al. In J Beha Nu Phys Ac (2022) 19:6 The sQ-HPF was based on he 14 ood g oups selec ed p e iously. Each i em asked abou he equency o con- sump ion o a pa icula ood g oup (Table4). Examples o ep esen a i e ood and be e age i ems included in each ood g oup we e p o ided o each i em. Es ima ed cu -o poin s we e used o de e mine he h eshold e- quency o consump ion o conside he esponden as an HPF consume o he i em, as shown in he column “C i- e ia o 1 poin ” in Table4. Fo he sel - epo ed e sion o he ques ionnai e, his column should be emo ed om he ques ionnai e o m since i is in ended o he use o he pe son assessing he sco e only. Using baseline da a om he PREDIMED-Plus FFQ, he pe cen age o HPF o e o al g ams pe day acco ding o he ques ionnai e i ems was calcula ed o each pa ien . This was used as he dependen a iable in a linea eg ession wi h he sQ-HPF sco e ob ained o each pa ien o es ablish he ollowing equa ion o he eg ession line: “HPF consump ion (% g/ day) = (3.7 x sQ-HPF sco e) + 7.6”. This equa ion allows he es ima ion o he HPF consump ion om he sQ-HPF sco e, as shown a he bo om o Table4. Weigh ed κ coe icien s we e calcula ed be ween HPF consump ion e iles acco ding o he sQ-HPF and he ou exis ing classi ica ion sys ems (Supplemen a y Table 2). The highes ag eemen was o he compa ison wi h UNC e iles (κ = 0.88), ollowed by IFIC e iles (κ = 0.65) and IARC e iles (κ = 0.50). The compa ison o e iles acco d- ing o he sQ-HPF wi h NOVA e iles o HPF showed a ai ag eemen (κ = 0.36). These compa isons we e in acco d- ance wi h he co esponding ag eemen plo s (Fig.2). Die a y, li es yle andca diome abolic cha ac e is ics o PREDIMED‑Plus pa icipan s acco ding o hesQ‑HPF sco e We aimed o in es iga e whe he changes in HPF con- sump ion as es ima ed om he FFQ and he ou di - e en sys ems we e also de ec ed when he sQ-HPF was Table 2 Associa ions be ween candida e sQ‑HPF i ems and e ile ag eemen a iable by binomial logis ic eg ession Binomial logis ic eg ession adjus ed o age, sex, ec ui men cen e , ene gy in ake, physical ac i i y le el, medica ion o blood p essu e and diabe es, wo king s a us, educa ional le el, and ci il s a us. Te ile ag eemen is he ou come a iable (sco e 0 – “low HPF consume ” i a subjec is classi ied in T1 o HPF consump ion by a leas wo classi ica ion sys ems; sco e 1 – “high HPF consume ” i a subjec is classi ied in T3 by a leas wo classi ica ion sys ems). Food g oups exp essed in equency o consump ion ( imes/day). HPF: highly p ocessed ood. I ems selec ed o he sQ‑HPF a e shown in bold. Bon : Bon e oni adjus men . N= 4400 P edic o a iable β SE p (Bon ) Fa y dai y p oduc s 0.33 0.10 0.024 Suga y dai y p oduc s 1.37 0.22 < 0.001 Cu ed mea 0.59 0.07 < 0.001 Fa s 0.63 0.14 0.001 Fe men ed alcohol 1.59 0.07 < 0.001 Dis illed alcohol 3.90 0.45 < 0.001 Suga y d inks 3.90 0.17 < 0.001 Swee s 0.46 0.05 < 0.001 Snacks 3.93 0.43 < 0.001 Ready To Ea (RTE) 2.91 0.33 < 0.001 Re ined ce eals 0.30 0.04 < 0.001 Sauces 0.97 0.18 < 0.001 Addi i es 0.10 0.02 < 0.001 F ied ood 0.84 0.15 < 0.001 Whole dai y p oduc s −0.01 0.06 1 Semi‑skimmed dai y p oduc s − 0.15 0.03 < 0.001 Eggs −0.31 0.19 1 Red mea 0.05 0.15 1 Whi e mea −1.41 0.17 < 0.001 Whi e ish −1.40 0.16 < 0.001 Blue ish −1.29 0.15 < 0.001 Vege ables −0.55 0.03 < 0.001 F ui −1.01 0.04 < 0.001 Po a oes −0.88 0.18 < 0.001 Nu s −0.87 0.07 < 0.001 Legumes −2.46 0.23 < 0.001 Oils −0.21 0.03 < 0.001 Non‑suga y d inks −0.09 0.03 0.302 Wholeg ain ce eals −0.53 0.04 < 0.001 Vi amin/supplemen s −0.10 0.12 1 Binge ea ing 0.17 0.14 1 N° binges/week 0.05 0.05 1 Snacking 0.06 0.08 1 Table 3 Explo a o y ac o analysis o sQ‑HPF i ems Ba le ’s es o sphe ici y =p< 0.001 Measu e o Sample Adequacy (MSA) = 0.78 S anda dized C onbach’s α = 0.67 N= 4400 Fac o 1: HPF die Fac o loadings Ready To Ea (RTE) 0.45 F ied ood 0.44 Fe men ed alcohol 0.42 Snacks 0.42 Fa y dai y p oduc s 0.38 Suga y d inks 0.37 Dis illed alcohol 0.36 Sauces 0.35 Re ined ce eals 0.35 Cu ed mea 0.34 Swee s 0.3 Suga y dai y p oduc s 0.29 Addi i es 0.26 Fa s 0.23 Page 8 o 16 Ma inez‑Pe eze al. In J Beha Nu Phys Ac (2022) 19:6 used. In addi ion, we wan ed o analyze he die a y p o- ile ac oss e iles o he sQ-HPF sco e, since his is a ool o e alua e a pa icula die a y pa e n. Die a y cha ac- e is ics o PREDIMED-Plus pa icipan s a baseline a e shown in Table5, g ouped by HPF consump ion e iles acco ding o he sco e ob ained in he sQ-HPF (T1 – lowes sco e, T3 – highes sco e). Consump ion o ood g oups included in he ques ionnai e was he highes o subjec s in T3 o he sQ-HPF sco e, which was no he case o ege ables, ui s, and legumes. Dai y p oduc s and ish consump ion we e no di e en among e iles. Wi h all he classi ica ion sys ems, subjec s in T3 showed he highes HPF consump ion (pe cen age o e o al g ams pe day: 12.45 ± 8.51 o NOVA, 54.85 ± 11.01 o IARC, 29.26 ± 12.02 o IFIC and 30.34 ± 12.40 o UNC, 35.05 ± 12.24 o ques ionnai e es ima ed HPF consump ion). Associa ions be ween li es yle and ca diome abolic pa ame e s and HPF consump ion ha e been p e iously epo ed, so we nex aimed o analyze i we could de ec changes in hese a iables ac oss e iles o he sQ-HPF sco e. Li es yle and ca diome abolic cha ac e is ics o PREDIMED-Plus pa icipan s a baseline g ouped by HPF consump ion e iles showed di e ences acco d- ing o he sQ-HPF sco e (Table6). Pa icipan s anked in he highes e ile had highe le els o iglyce ides (161.05 ± 90.55 mg/dL), highe weigh (90.68 ± 13.10 kg) and wais ci cum e ence (109.83 ± 9.45 cm) compa ed o hose in he lowes e ile. No signi ican changes among e iles we e de ec ed in as ing glucose and o al cho- les e ol le els. Subjec s in T3 o he sQ-HPF sco e spen mo e ime wa ching TV han T1 subjec s (4.05 ± 2.02 h/ day), while sleeping hou s we e simila ac oss e iles. A ound hal o he subjec s classi ied in T3 we e clas- si ied as seden a y (51.5%). Conce ning ea ing habi s, 30.5% o he subjec s wi h he highes sQ-HPF sco e epo ed snacking. Assessing longi udinal changes inHPF consump ion wi h hesQ‑HPF Longi udinal analysis o HPF consump ion es ima ed om he FFQ by each classi ica ion sys em e ealed sig- ni ican di e ences ac oss he i s 3 yea s o he PRED- IMED-Plus s udy. In all cases, mean HPF consump ion showed a end owa ds a dec ease ha was lowe in yea 2 han baseline (Table7). This was also he case when Table 4 Sc eening ques ionnai e o highly p ocessed ood consump ion (sQ‑HPF) Wha was you a e age equency o consump ion o e he pas yea o … C i e ia o 1 poin Sco e (0 o 1) Q1. Fa y dai y p oduc s (c eam, cu ed o semi‑cu ed cheese, p ocessed so cheese wedges)? > = 2 /week Q2. Suga y dai y p oduc s (condensed milk, indus ially p oduced milkshakes, la o ed Pe i Suisse yogu , cus a d, c ème ca amel lan, pudding)? > 3 /mon h Q3. Cu ed mea s (se ano ham, sandwich (deli) ham, cu ed cold mea s, pâ é, bacon, ma bling)? > = 1 /day Q4. Fa s (ma ga ine, bu e , la d (animal a ))? > 3 /mon h Q5. Fe men ed alcohols ( osé wine, musca el wine, young ed wine, aged ed wine, whi e wine, Spanish spa kling wine (ca a), bee )? > 1 /day Q6. Dis illed alcohols (liquo s, anise e, whisky, gin, odka, cognac)? > 3 /mon h Q7. Suga y and a i icially swee ened d inks (so d inks, a i icially swee ened d inks, bo led juice, g ape mus )? > = 2 /week Q8. Swee s (ice‑c eam and so be s, canned ui in juice o sy up, biscui s, whole meal biscui s, chocola e biscui s, honey, homemade baking p oduc s, indus ially p oduced con ec ione y, donu s, mu ins, cupcakes, indus ially p oduced cakes, chu os, chocola es, soluble cocoa powde , nouga , ma zipan, sho b ead biscui s, jam)? > 1 /day Q9. Snacks (packaged po a o c isps/chips, packaged snacks)? > 3 /mon h Q10. Ready To Ea p oduc s (pizza, c oque es, ins an soup)? > 3 /mon h Q11. Re ined ce eals (whi e and sliced b ead, b eak as ce eals, Spaghe i, maca oni, Spanish noodles, whi e ice)? > = 2 /week Q12. Sauces (mus a d, mayonnaise, oma o sauce, Ke chup)? > 1 /week Q13. Addi i es (suga , able sal )? > 3 /day Q14. F ied oods (ea ou and homemade)? > = 2 /week TOTAL SCORE: Equi alency be ween sQ‑HPF sco e and he es ima ed pe cen age o HPF consump ion o e he o al in ake in g ams pe day: Sco e 1 2 3 4 5 6 7 8 9 10 11 12 13 14 % HPF 11.3 15 18.7 22.4 26.1 29.8 33.5 37.2 40.9 44.6 48.3 52 55.7 59.4 Page 9 o 16 Ma inez‑Pe eze al. In J Beha Nu Phys Ac (2022) 19:6 HPF consump ion was es ima ed h ough he sQ-HPF (22.9 ± 0.22% o g/day a baseline and 17.3 ± 0.22% o g/ day in yea 2). Discussion This a icle p esen s he sQ-HPF, a sho , in eg a i e, and easy- o-use ques ionnai e o es ima e HPF consump ion (Table4). The de elopmen p ocess in ol ed a ca e ully designed s a is ical analysis and p ac ical conside a ions o i s use in a clinical/epidemiological con ex . This new ool includes 14 ood and be e age i ems o which he equency o consump ion is eco ded, based on a p e iously alida ed FFQ om he PREDIMED-Plus T ial [38–40]. Each i em is sco ed as 1 i he equency o consump ion co esponds o he HPF die a y pa e n, acco ding o he calcula ed h esholds, and as 0 o he - wise. The e o e, he highe he sco e, he highe he con- sump ion o HPF. Mo eo e , his sco e can be used o es ima e he pe cen age o HPF consump ion o e o al in ake, a oiding he need o adminis e a leng hy FFQ (Table4) [63]. The p esen wo k demons a es ha s a is ical app oaches such as EFA and C onbach’s analysis a e aluable ools o de eloping in eg a i e ools ela ed o die a y pa e ns and ea ing habi s. Indeed, analysis o ques ionnai e i ems h ough EFA demons a ed ha hey iden i ied one co e cons uc , he HPF die a y pa - e n (Table3) [64]. This was expec ed conside ing ha i ems we e selec ed based on a posi i e associa ion wi h he a iable e ile ag eemen h ough binomial logis- ic eg ession (Table 1). This esul ed in 14 ques ions based on he equency o consump ion o ood g oups associa ed wi h a highe HPF consump ion. One o he easons o his was ha , in his way, he ques ionnai e could iden i y people wi h an HPF die a y pa e n wi h a ocus on he equency o consump ion and no on he speci ic HPF i ems hey consumed, so ha he ques- ionnai e could de ec di e en HPF die a y pa e ns. Fig. 2 Ag eemen plo s o e iles o HPF consump ion. Visual ep esen a ion o con ingency ables be ween he sQ‑HPF e iles and A NOVA HPF e iles, B IARC HPF e iles, C IFIC HPF e iles and D UNC HPF e iles. Ma ginal o als o he con ingency able a e loca ed on he op and igh axis. Shading ep esen s he le el o ag eemen , black indica es “pe ec ag eemen ”, and g ey indica es “pa ial ag eemen ”. The ex en o which ec angles de ia e om he diagonal line o no bias indica es he ex en o disag eemen , and he posi ion (abo e/below) indica es di ec ion o he disag eemen . HPF: highly p ocessed ood Page 16 o 16 Ma inez‑Pe eze al. In J Beha Nu Phys Ac (2022) 19:6 • as , con enien online submission • ho ough pee e iew by expe ienced esea che s in you ield • apid publica ion on accep ance • suppo o esea ch da a, including la ge and complex da a ypes • gold Open Access which os e s wide collabo a ion and inc eased ci a ions maximum isibili y o you esea ch: o e 100M websi e iews pe yea • A BMC, esea ch is always in p og ess. Lea n mo e biomedcen al.com/submissions Ready o submi you esea ch Ready o submi you esea ch ? Choose BMC and bene i om: ? Choose BMC and bene i om: 62. C onbach LJ. Coe icien alpha and he in e nal s uc u e o es s. Psycho‑ me ika. 1951;16(3):297–334 [ci ed 2021 Feb 22]. A ailable om: h ps:// link. sp in ge . com/ a ic le/ 10. 1007/ BF023 10555. 63. Fangupo LJ, Hasza d JJ, Leong C, Hea h ALM, Fleming EA, Taylo RW. Rela i e alidi y and ep oducibili y o a ood equency ques ionnai e o assess ene gy in ake om minimally p ocessed and ul a‑p ocessed oods in young child en. Nu ien s. 2019;11(6):1–13. 64. Koslowsky M, Scheinbe g Z, Bleich A, Ma k M, Ap e A, Danon Y, e al. The ac o s uc u e and c i e ion alidi y o he sho o m o he ea ing a i udes es . J Pe s Assess. 1992;58(1):27–35 [ci ed 2021 May 14]. A ail‑ able om: h ps:// www. and online. com/ doi/ abs/ 10. 1207/ s1532 7752j pa5801_3. 65. Gibney MJ, Fo de CG, Mullally D, Gibney ER. Ul a‑p ocessed oods in human heal h: a c i ical app aisal. Am J Clin Nu . 2017;106(3):717–24 [ci ed 2020 Jun 25]. A ailable om: h p:// ajcn. nu i ion. o g/ lookup/ doi/ 10. 3945/ ajcn. 117. 160440. 66. Nieuwenhuijsen MJ. Design o exposu e ques ionnai es o epidemiologi‑ cal s udies. Occup En i on Med. 2005;62(4):272–80 [ci ed 2021 May 11]. A ailable om: h p:// oem. bmj. com/. 67. San‑C is obal R, Na as‑Ca e e o S, Celis‑Mo ales C, B ennan L, Walsh M, Lo eg o e JA, e al. Analysis o die a y pa e n impac on weigh s a us o pe sonalised nu i ion h ough on‑line ad ice: he ood4Me Spanish coho . Nu ien s. 2015;7(11):9523–37. 68. San‑C is obal R, Na as‑Ca e e o S, Celis‑Mo ales C, Li ings one KM, S ew‑ a ‑Knox B, Rankin A, e al. Cap u ing heal h and ea ing s a us h ough a nu i ional pe cep ion sc eening ques ionnai e (NPSQ9) in a andomised in e ne ‑based pe sonalised nu i ion in e en ion: he Food4Me s udy. In J Beha Nu Phys Ac . 2017;14(1):1–12. 69. T ichopoulou A, Cos acou T, Bamia C, T ichopoulos D. Adhe ence o a Medi e anean die and su i al in a G eek popula ion. N Engl J Med. 2003;348(26):2599–608 [ci ed 2021 May 10]. A ailable om: h ps:// pubmed. ncbi. nlm. nih. go / 12826 634/. 70. Te sunen H‑M, Han unen S, Tuomainen T‑P, Vi anen JK. Adhe ence o a heal hy No dic die and isk o ype 2 diabe es among men: he Kuopio Ischaemic hea disease isk ac o s udy. Eu J Nu . 2021; [ci ed 2021 May 10];Online pub. A ailable om: h ps:// link. sp in ge . com/ 10. 1007/ s00394‑ 021‑ 02569‑1. 71. Toh DWK, Xia X, Su an o CN, Low JHM, Poh KK, Wang J‑W, e al. Enhanc‑ ing he ca dio ascula p o ec i e e ec s o a heal hy die a y pa e n wi h wol be y (Lycium ba ba um): a andomized con olled ial. Am J Clin Nu . 2021; [ci ed 2021 May 10];Online ahe. A ailable om: h ps:// acade mic. oup. com/ ajcn/ ad an ce‑ a ic le/ doi/ 10. 1093/ ajcn/ nqab0 62/ 62726 07. 72. Pa e son EN, Ne ille CE, Wallace SM, Woodside JV, Kee F, Young IS, e al. Die a y pa e ns associa ed wi h enal impai men in he No he n I eland coho o he longi udinal s udy o ageing (NICOLA). Eu J Nu . 2021; [ci ed 2021 May 10];Published. A ailable om: h ps:// link. sp in ge . com/ 10. 1007/ s00394‑ 021‑ 02579‑z. 73. Guen he PM, Ki kpa ick SI, Reedy J, K ebs‑Smi h SM, Buckman DW, Dodd KW, e al. The heal hy ea ing Index‑2010 is a alid and eliable measu e o die quali y acco ding o he 2010 die a y guidelines o Ame icans 1‑3. J Nu Me hodol Ma h Model J Nu . 2014;144:399–407 [ci ed 2021 May 10]. A ailable om: h p:// jn. nu i ion. o g. 74. K ebs‑Smi h SM, Pannucci TRE, Suba AF, Ki kpa ick SI, Le man JL, Tooze JA, e al. Upda e o he heal hy ea ing index: HEI‑2015. J Acad Nu Die . 2018;118(9):1591–602 [ci ed 2021 May 14]. A ailable om: h ps:// pub‑ med. ncbi. nlm. nih. go / 30146 071/. 75. McCullough ML, Wille WC. E alua ing adhe ence o ecommended die s in adul s: he Al e na e Heal hy Ea ing Index. Public Heal h Nu . 2006;9(1A):152–7 [ci ed 2021 May 14]. A ailable om: h ps:// pubmed. ncbi. nlm. nih. go / 16512 963/. 76. Food S anda ds Sco land. B ie ing pape on disc e iona y oods. 2018. 77. Na ional Heal h and Medical Resea ch Council. Aus alian die a y guide‑ lines. Canbe a; 2013. [ci ed 2021 May 10]. A ailable om: h p:// www. nhm c. go . au 78. G iege JA, Wyche ley TP, Johnson BJ, Golley RK. Disc e e s a egies o educe in ake o disc e iona y ood choices: a scoping e iew. In J Beha Nu Phys Ac . 2016;13(1):1–22 [ci ed 2021 May 10]. A ailable om: h ps:// ijbnpa. biome dcen al. com/ a ic les/ 10. 1186/ s12966‑ 016‑ 0380‑z. 79. Li ings one KM, Celis‑Mo ales C, Na as‑Ca e e o S, San‑C is obal R, Fo s e H, Woolhead C, e al. Pe sonalised nu i ion ad ice educes in ake o disc e iona y oods and be e ages: indings om he Food4Me andomised con olled ial. In J Beha Nu Phys Ac . 2021;18(1):70 [ci ed 2021 Jun 8]. A ailable om: h ps:// ijbnpa. biome dcen al. com/ a ic les/ 10. 1186/ s12966‑ 021‑ 01136‑5. 80. Phulke d S, Law ence M, Vande ij e e S, Sacks G, Wo sley A, Tangcha oen‑ sa hien V. A e iew o me hods and ools o assess he implemen a ion o go e nmen policies o c ea e heal hy ood en i onmen s o p e en ing obesi y and die ‑ ela ed non‑communicable diseases. Implemen Sci. 2016;11(15) [ci ed 2021 May 11]. A ailable om: h ps:// pubmed. ncbi. nlm. nih. go / 26846 789/. 81. Sa bagili‑Shaba C, Zelbe ‑Sagi S, Fliss Isako N, Ron Y, Hi sch A, Maha shak N. De elopmen and alida ion o p ocessed oods ques ion‑ nai e (PFQ) in adul in lamma o y bowel diseases pa ien s. Eu J Clin Nu . 2020;74(12):1653–60 A ailable om: h ps:// doi. o g/ 10. 1038/ s41430‑ 020‑ 0632‑5. 82. Mo a VW de L, Lima SCVC, Ma chioni DML, Ly a C de O. Food equency ques ionnai e o adul s in he B azilian no heas egion: emphasis on he le el o ood p ocessing. Re Saude Publica 2021;55(51). [ci ed 2021 Aug 24]. A ailable om: h ps:// www. e is as. usp. b / sp/ a ic le/ iew/ 189620. Publishe ’s No e Sp inge Na u e emains neu al wi h ega d o ju isdic ional claims in pub‑ lished maps and ins i u ional a ilia ions.