Does anthropometric status at 6 months predict the over-dispersion of malaria infections in children aged 6–18 months? A prospective cohort study

Bendabendae al. Mala J (2019) 18:143 h ps://doi.o g/10.1186/ s12936-019-2778-y RESEARCH Does an h opome ic s a us a 6mon hs p edic heo e -dispe sion o mala ia in ec ions inchild en aged 6–18mon hs? Ap ospec i e coho s udy Jaden Bendabenda1,2, Noel Pa son1,4, Lo a Hallamaa2, Ulla Asho n2, Ka h yn G. Dewey3, Pe Asho n2 and Kenne h Male a1* Abs ac Backg ound: In mala ia-endemic se ings, a small p opo ion o child en su e epea ed mala ia in ec ions, con ib- u ing o mos o he mala ia cases, ye unde lying ac o s a e no ully unde s ood. This s udy was aimed o de e mine whe he unde nu i ion p edic s his o e -dispe sion o mala ia in ec ions in child en aged 6–18 mon hs in se ings o high mala ia and unde nu i ion p e alence. Me hods: P ospec i e coho s udy, conduc ed in Mangochi, Malawi. Six-mon hs-old in an s we e en olled and had leng h- o -age z-sco es (LAZ), weigh - o -age z-sco es (WAZ), and weigh - o -leng h z-sco es (WLZ) assessed. Da a we e collec ed o ‘p esumed’, clinical, and apid diagnos ic es (RDT)-con i med mala ia un il 18 mon hs. Mala ia mic oscopy was done a 6 and 18 mon hs. Nega i e binomial eg ession was used o mala ia incidence and modi ied Poisson eg ession o mala ia p e alence. Resul s: O he 2723 child en en olled, 2561 (94%) had an h opome y and mala ia da a. The mean (s anda d de ia- ion [SD]) o LAZ, WAZ, and WLZ a 6 mon hs we e − 1.4 (1.1), − 0.7 (1.2), and 0.3 (1.1), espec i ely. The mean (SD) incidences o ‘p esumed’, clinical, and RDT-con i med mala ia om 6 o 18 mon hs we e: 1.1 (1.6), 0.4 (0.8), and 1.3 (2.0) episodes/yea , espec i ely. P e alence o mala ia pa asi aemia was 4.8% a 6 mon hs and 9.6% a 18 mon hs. Highe WLZ a 6 mon hs was associa ed wi h lowe p e alence o mala ia pa asi aemia a 18 mon hs (p e alence a io [PR] = 0.80, 95% con idence in e al [CI] 0.67 o 0.94, p = 0.007), bu no wi h incidences o ‘p esumed’ mala ia (inci- dence a e a io [IRR] = 0.97, 95% CI 0.92 o 1.02, p = 0.190), clinical mala ia (IRR = 1.03, 95% CI 0.94 o 1.12, p = 0.571), RDT-con i med mala ia (IRR = 1.00, 95% CI 0.94 o 1.06, p = 0.950). LAZ and WAZ a 6 mon hs we e no associa ed wi h mala ia ou comes. Household asse s, ma e nal educa ion, and ood insecu i y we e signi ican ly associa ed wi h mala ia. The e we e signi ican a ia ions in hospi al-diagnosed mala ia by s udy si e. Conclusion: In child en aged 6–18 mon hs li ing in mala ia-endemic se ings, LAZ, WAZ, and WLZ do no p edic mala ia incidence. Howe e , WLZ may be associa ed wi h p e alence o mala ia. Socio-economic and mic o-geo- g aphic ac o s may explain he a ia ions in mala ia, bu hese equi e u he s udy. T ial egis a ion NCT00945698. Regis e ed July 24, 2009, h ps ://clini cal ials.go /c 2/show/NCT00 94569 8, NCT01239693. Regis e ed No 11, 2010, h ps ://clini cal ials.go /c 2/show/NCT01 23969 3 Keywo ds: Child en, iLiNS s udies, In ec ions, Mala ia, O e -dispe sion, S un ing, Unde nu i ion, Was ing © The Au ho (s) 2019. This a icle is dis ibu ed unde he e ms o he C ea i e Commons A ibu ion 4.0 In e na ional License (h p://c ea i ecommons.o g/licenses/by/4.0/), which pe mi s un es ic ed use, dis ibu ion, and ep oduc ion in any medium, p o ided 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 license, and indica e i changes we e made. The C ea i e Commons Public Domain Dedica ion wai e (h p://c ea i ecommons.o g/ publicdomain/ze o/1.0/) applies o he da a made a ailable in his a icle, unless o he wise s a ed. Open Access Mala ia Jou nal *Co espondence: [email p o ec ed] 1 Depa men o Public Heal h, School o Public Heal h, Uni e si y o Malawi College o Medicine, Maha ma Gandhi Road, P i a e Bag 360, Blan y e 3, Malawi Full lis o au ho in o ma ion is a ailable a he end o he a icle Page 2 o 12 Bendabendae al. Mala J (2019) 18:143 Backg ound Mala ia is one o he mos se ious public heal h p oblems in he wo ld, wi h an es ima ed 216 million cases and 445,000 dea hs epo ed in 2016, and app oxima ely 90% o he cases and dea hs occu ing in he A ican Region [1]. Malawi is one o he mala ia-hype endemic coun ies in sub-Saha an A ica, wi h a ound 4 million mala ia cases epo ed annually [2]. In mala ia-hype endemic coun ies, i is assumed ha i ually all exposed indi iduals would su e a mala ia episode by ea ly childhood [3]. Howe e , se e al s udies ha e shown ha in hese se ings, only a small p opo ion o child en su e epea ed mala ia in ec ions, and hese child en a e esponsible o mos o he mala ia cases [4–6], an o e -dispe sion known as he ‘20/80’ ule [7]. The sugges ed unde lying ac o s o his o e -dis- pe sion in mala ia in ec ions a e a ied, and include o ins ance gene ic [5], beha iou al [8] and en i onmen- al ac o s [6]. O he s udies sugges ha unde nu i- ion plays an impo an ole in mala ia epidemiology because o he syne gis ic in e ac ions be ween nu i- ion and in ec ions [9–11]. Howe e , se e al e iews on he ela ionship be ween unde nu i ion and mala ia de e mined ha he cu en e idence is inconclusi e; a ibu ed o he he e ogenei y in he s udy popula ions, mala ia pa asi e species, and hos -pa asi e ela ionship [12–14]. Hence he need o u he unde s anding o he ole o unde nu i ion in mala ia epidemiology. The In e na ional Lipid-based Nu ien Supplemen s (iLiNS) P ojec DOSE and DYAD s udies we e and- omized con olled ials conduc ed in Malawi o s udy he impac o lipid-based nu ien supplemen s (LNS) on g ow h o child en [15, 16]. Analysis o longi udi- nal mala ia da a in he wo s udies showed ha 39% o he in an s and young child en aged 6 o 18mon hs did no epo any mala ia episode in he one-yea s udy pe iod; only 30.7% epo ed mo e han one epi- sode o ‘p esumed’ mala ia bu hese we e esponsible o 73.7% o he ‘p esumed’ mala ia episodes [17]. The aim o he cu en analysis was o u he in es iga e he dis ibu ion o mala ia in child en in hese coho s and de e mine whe he his dis ibu ion is p edic ed by an h opome ic s a us a 6mon hs. The hypo hesis was ha lowe leng h o age z-sco es (LAZ), weigh - o -age z-sco es (WAZ), and lowe weigh o leng h z-sco es (WLZ) a 6mon hs will be associa ed wi h a highe incidence o mala ia om 6 o 18mon hs, and highe p e alence o mala ia a 18mon hs. Me hods S udy se ing The iLiNS-DOSE and iLiNS-DYAD-M s udies we e conduc ed in ou acili ies: one public dis ic hospi al (Mangochi), one mission hospi al (Malindi), and wo u al public heal h cen es (Lungwena and Namwe a) in Mangochi Dis ic , Sou he n Malawi. The o al ca ch- men popula ion o 180,000 la gely subsis ed on a m- ing and ishing. Mangochi si e is low-lying a an al i ude o ~ 485 m abo e sea le el, bu a e sed by he Shi e Ri e ( he la ges i e in Malawi). Two o he s udy si es (Lungwena and Malindi) a e also low-lying wi h he simila al i ude along he eas e n sho e o Lake Malawi. In con as , Namwe a lies a he op o Namwe a Hills, bo de ing Mozambique, a an al i ude o ~ 900m abo e sea le el and is a om he la ge wa e bodies. Namwe a expe ienced highe ain all and coole empe a u es han he o he h ee s udy si es [18, 19]. In Malawian child en aged < 5yea s, he p e alence o mala ia (by mic oscopy), dia hoea and acu e espi a o y in ec ions we e 24.3%, 22% and 5%, espec i ely, wi h seasonal luc ua ions [2, 20]. The sub- opical clima e comp ising a wa m, we season om No embe o Ap il, a cool, d y win e season om May o Augus , and a ho , d y season om Sep embe o Oc obe [21] is a ou - able o he Anopheles mosqui oes which ansmi Plas- modium pa asi es. Plasmodium alcipa um is he mos dominan and causes abou 98% o all mala ia in ec ions in Malawi. Mala ia ansmission occu s h oughou he yea wi h highes ansmission a es occu ing be ween Oc obe and Ap il ( ainy season), mainly in low-lying and high empe a u e a eas. S udy design, da a collec ion ande hics s a emen The da a o his analysis we e aken om he iLiNS- DOSE and iLiNS-DYAD-M s udies— wo la ge commu- ni y-based andomized con olled ials conduc ed in u al Malawi. In he iLiNS-DOSE s udy, 6-mon hs old child en we e andomly alloca ed o one o i e in e en ion g oups p o ided wi h di e en doses o o mula ions o LNS o o a con ol g oup ha did no ecei e LNS du ing he 12-mon h s udy pe iod, be ween No embe 2009 and May 2012. In he iLiNS-DYAD-M s udy, p egnan women < 20weeks’ ges a ion we e andomly alloca ed o one o h ee g oups o ecei e i on and olic acid (IFA), mul iple mic onu ien s (MMN) o a small-quan i y (20g) o LNS daily. A e deli e y, women in he IFA g oup ecei ed placebo able s, while MMN and LNS supplemen a ion was con inued up o 6mon hs pos pa - um. Child en o mo he s in he LNS g oup also ecei ed LNS 10g wice daily om 6 o 18mon hs. This s udy was conduc ed om Feb ua y 2011 o Ap il 2015. De ails o design, andomiza ion and en olmen o he wo s udies can be ound in he main ou come pape s [15, 16]. In bo h s udies, esea ch assis an s isi ed he child en’s homes e e y week om age 6 o 18mon hs o in e iew Page 3 o 12 Bendabendae al. Mala J (2019) 18:143 he gua dians abou he child’s heal h in he p e ious 7days using a s uc u ed ques ionnai e. The in o ma ion was complemen ed by a pic u e calenda illed ou by he gua dians daily o aid memo y o hei child en’s mo - bidi y s a us. These we e done o minimise p oblems o ecall associa ed wi h communi y mo bidi y assessmen s [22, 23]. The use o ma e nal in e iews o collec da a on child mo bidi y has been alida ed in p e ious s ud- ies [24, 25]. The esea ch assis an s e e ed all cases o e e o he nea by heal h acili y o a mala ia apid diag- nos ic es (RDT) and ea men wi h a eme he /lume- an ine, he na ionally ecommended an i-mala ial d ug. The child en we e ollowed h oughou he yea , co e ing pe iods o bo h high and low mala ia ansmission. Facili y heal h wo ke s we e ained o collec da a on clinical diagnosis, RDT, and/o mala ia mic oscopy esul s whene e he child was ea ed a he heal h acili y. In addi ion, all child en had mala ia mic oscopy es s done du ing he scheduled s udy clinic isi s a age 6mon hs and 18mon hs. Blood smea s we e ob ained om all child en a he ime o he blood sampling o biochemical assessmen s, s ained wi h 2% Giemsa o 30min. All hick slides we e e iewed by wo mic osco- pis s using a high-powe mic oscope o de e mine he p esence o mala ia pa asi es. Disc epan eadings we e esol ed by a hi d e iewe . Mala ia RDT was also done a 6mon hs scheduled clinic isi using apid es ki (Clea iew Mala ial Combo, Ale e, Sou h A ica). An h opome ic assessmen s we e done a 6mon hs and 18mon hs. S udy an h opome is s measu ed he in an ’s leng h wi h a high-quali y leng h boa d (Ha penden In an- ome e ; Hol ain Limi ed) and eco ded i o he nea es 1 mm. They weighed unclo hed in an s wi h elec onic in an weighing scale (SECA 735; Seca GmbH & Co), eco ding o he nea es 10g. The an h opome is s we e ained and hei measu emen eliabili y was e i ied a he s a o he s udy and a 6-mon hs in e als he e- a e wi h me hods adap ed om he p ocedu es used in he WHO Mul icen e G ow h Re e ence S udy [26]. The an h opome is s calib a ed all equipmen wi h s anda d weigh s and leng h ods daily. S udy nu ses collec ed 5–7mL o blood by enepunc- u e using a 23-gauge needle in o 7.5mL e acua ed, ace elemen - ee polye hylene ubes con aining li hium hep- a in (Sa s ed Mono e e, NH4‐hepa in, Sa s ed Inc., New on, NC, USA). The blood ube was immedia ely in e ed 10 imes o mix he hepa in an icoagulan wi h he blood o p e en clo ing. A small aliquo o he whole blood was pipe ed ou and used o analyse haemoglobin (Hb) on he Hemocue 201+ sys em (Hemocue, B ea, CA, USA). The ube con aining he emaining whole blood was hen placed in an insula ed coole wi h ice packs and p ocessed wi hin 2h o collec ion. T ained labo a o y s a hen aliquo ed whole blood in o mic ocu e es and measu ed zinc p o opo phy in (ZPP) concen a ion om unwashed enous blood sample using a haema o luo om- e e (206D, AVIV Biomedical Inc., Lakewood, NJ, USA). A en olmen , esea ch assis an s in e iewed he mo he s o ob ain household le el in o ma ion includ- ing asse s, numbe o child en, ma e nal educa ion (yea s spen in school) and ma e nal age. Household ood secu i y was assessed using Household Food Insecu i y Access Scale (HFIAS) [27]. Use o insec icide ea ed bed ne s (ITNs) was assessed by asking he gua dians he numbe o days ha he child slep unde a bed ne du - ing he week p eceding he in e iew a 6mon hs. Si e was de ined based on he clinics whe e he wo s udies we e conduc ed (i.e. Namwe a, Mangochi, Malindi, and Lungwena). De ini ion o  hep edic o s and heou comes Ou come a iables The p ima y ou come o his analysis was he incidence o ‘p esumed’ mala ia, de i ed om he weekly mo bidi y da a. ‘P esumed’ mala ia diagnosis was de ined as his o y o e e ei he epo ed by he gua dians o ympanic empe a u e ≥ 38°C measu ed by he esea ch assis an s du ing he home isi . An episode o ‘p esumed’ mala ia was de ined as he pe iod s a ing om he day he child had mala ia symp oms, p eceded by a leas 2days o no symp oms o no da a. The episode ended on he las day he child had mala ia symp oms, i ollowed by a leas 2 symp om- ee days. Fe e episodes accompanied by espi a o y signs o dia hoea we e excluded om he mala ia diagnosis. Seconda y ou comes included (1) incidence o clinical mala ia, aken om he mala ia diagnosis made by he heal h wo ke in he absence o a diagnos ic es when- e e he child isi ed a heal h acili y, (2) incidence o con i med mala ia, aken om he mala ia diagnosis made by he heal h wo ke con i med by a posi i e RDT, and (3) p e alence o mala ia pa asi aemia de i ed om he mala ia mic oscopy esul s a age 18mon hs. Mala ia incidence was calcula ed as o al episodes o mala ia/ o al child yea s a isk; mala ia pa asi aemia p e alence was calcula ed as he p opo ion wi h a posi i e mala ia pa asi e slide. P edic o a iables Leng h o age z-sco es (LAZ), weigh - o -age z-sco es (WAZ), and weigh o leng h z-sco es (WLZ) we e calcula ed om he an h opome y da a using WHO g ow h s anda ds [26]. I on de iciency a age 6mon hs was de ined as whole blood ZPP > 70 µmol/mol haem [28]. HFIAS z-sco es we e gene a ed by summing he Page 4 o 12 Bendabendae al. Mala J (2019) 18:143 alue o esponses o nine ques ions ega ding ood inse- cu i y: he highe he sco e, he highe deg ee o ood insecu i y in he las 4weeks [27]. Household asse sco es we e de ined as he p incipal componen s sco e based on baseline owne ship o a se o asse s and household qual- i y: he highe he sco e, he be e he li ing condi ions. Numbe o child en aged < 5yea s was de ined as num- be o child en below he age o i e who we e pa o he pa icipan ’s household a 6mon hs. S a is ical analysis All child en who had mala ia da a a any poin om age 6 o 18mon hs we e included in he analysis. Nega i e binomial eg ession was used o assess he associa ion o LAZ, WAZ, and WLZ (independen a iables) wi h he incidence o mala ia (dependen a iable), and Pois- son eg ession (wi h a obus a iance es ima o ) [29] o assess he associa ion o LAZ, WAZ, and WLZ wi h p e alence o mala ia pa asi aemia. To s udy he independen e ec o a ious p edic o s, mul i a ia e eg ession models we e cons uc ed ha included po en ial p edic o s collec ed a 6mon hs. The ollowing we e po en ial p edic o s: household asse sco es, ma e nal age (cen ed a ound he mean), and edu- ca ion, numbe o child en aged < 5yea s, HFIA sco e, whe he he child ecei ed he s udy in e en ion (LNS) o no , adjus ed o he addi ional con ol g oup [MMN] in iLiNS-DYAD, sex o he child, Hb concen a ion, i on s a us, mon h o bi h, daily use o ITNs, dis ance o heal h acili y and s udy si e. Collinea i y among he p e- dic o s was es ed using he collin s a a command. I he p edic o s we e highly collinea (> 0.5), he one ha was less s ongly associa ed wi h he ou comes was d opped. The esul s a e epo ed as incidence a e a ios [IRR] o p e alence a io [PR] and hei 95% con idence in e - als (95% CI) a p = 0.05. Robus s anda d e o s we e compu ed o adjus o co ela ion o ecu en mala ia episodes in a single child. O he po en ial p edic o s we e conside ed including immuniza ion s a us, ma ke s o in lamma ion (C- eac i e p o ein and alpha1-acid glycop o ein concen a ions) a 6mon hs, ma e nal and child mala ia immuni y, and ma e - nal HIV s a us. Howe e , hese a iables we e a ailable only om a subsample o he wo s udies, hence we e e en u- ally d opped om he inal models o maximize he sample size. Fu he mo e, hese a iables showed li le e ec on he model du ing sensi i i y analysis. S a a e sion 14 (S a a- Co p, Texas, USA) was used o he main analyses. Resul s S udy popula ion O he 2723 child en en olled in he wo s udy coho s, 2561 (94%) had da a o he p ima y ou come (1928 child en om he iLiNS DOSE s udy and 633 child en om he iLiNS DYAD-M s udy). These we e included in he inal analysis (Fig.1). A age 6mon hs, he mean (SD) leng h- o -age z-sco es (LAZ), weigh - o -age z-sco es (WAZ), and weigh - o -leng h z-sco es (WLZ) we e − 1.4 (1.1), − 0.7 (1.2), and 0.3 (1.1) espec i ely. The p o- po ions o child en who we e s un ed, unde weigh and was ed we e 28.3%, 12.9%, and 2.0%, espec i ely. Fu he cha ac e is ics o hese child en a age 6mon hs a e sum- ma ized in Table1. Incidence andp e alence o mala ia The child en con ibu ed 2405.6 child yea s o ollow up, i.e. he mean (SD) du a ion o ollow up was 344 (73) days/child. Du ing he home isi s, 27,340 mo - bidi y episodes we e epo ed, 9.3% (2549/27,340) o which we e episodes o ‘p esumed’ mala ia. The es we e due o: ARI, 53.8% (14,708/27,340); dia - hoea, 23% (6282/27,340) and mino condi ions, 13.9% (3801/27,340). O he o al mo bidi y episodes iden i ied a he home isi s, 44% (12,048/27,340) we e epo ed and ea ed a a heal h acili y, 32.5% (3917/12,048) o which we e ea ed o mala ia, 76.8% (3007/3917) o hem con i med by RDT. F om he home isi s da a, he mean (SD) incidence o all illnesses combined was 16.2 (13.1) episodes pe child yea . The mean (SD) incidence o ‘p esumed’ mala ia was 1.1 (1.6) episodes pe child yea . The mean (SD) incidence o acu e espi a o y in ec ions was 7.8 (8.0) episodes pe child yea and he mean (SD) incidence o dia hoea was 2.7 (2.7) episodes pe child yea . Con- e sely, he na ional da a show ha dia hoea incidence is highe han ARI incidence in child en unde - i e, p obably because he age anges a e sligh ly di e en . The na ional da a do no p o ide speci ic incidences o child en unde 18mon hs. F om he hospi al isi s da a, he mean (SD) incidences o clinical mala ia and con i med mala ia we e, espec- i ely, 0.4 (0.8) and 1.3 (2.0) episodes pe child yea (i.e. he mean (SD) incidence o all mala ia a hospi al isi s was 1.7 (2.4) episodes pe child yea ). The p e alence o mala ia pa asi aemia (by mic os- copy) inc eased om 4.8% a age 6mon hs o 9.6% a age 18 mon hs. Du ing he 12-mon h ollow up pe iod, 45.1% (1156/2561) o he child en included in his analysis did no epo any episode o mala ia, 28.4% (728/2561) epo ed one episode and 26.4% (677/2561) epo ed > 1 mala ia episodes. The child en who epo ed > 1 mala ia episodes we e esponsible o 71.4% (1821/2549) o all mala ia episodes epo ed in he wo s udies. These indings we e simila ac oss he di e en de ini ions o mala ia. Page 5 o 12 Bendabendae al. Mala J (2019) 18:143 Associa ion o LAZ a 6mon hs wi hmala ia om6 o18mon hs In bi a ia e analysis, a highe LAZ a 6mon hs was associa ed wi h a lowe incidence o clinical mala ia and lowe incidence o con i med mala ia om 6 o 18 mon hs (Table 2, unadjus ed), and lowe p e a- lence o mala ia pa asi aemia a 18mon hs (Table3, unadjus ed), bu no incidence o ‘p esumed’ mala ia (Table2, unadjus ed). When adjus ed o o he p e- dic o s, hese associa ions we e no longe e iden (Tables2, 3, adjus ed, and 4). Associa ion o WLZ a 6mon hs wi hmala ia om6 o18mon hs The e was no associa ion be ween WLZ a 6mon hs and he incidence o ‘p esumed’ mala ia, clinical mala ia o con i med mala ia om 6 o 18mon hs, in ei he bi a i- a e o mul i a ia e analyses (Table 2). In mul i a ia e analysis, highe WLZ a 6mon hs was associa ed wi h lowe p e alence o mala ia pa asi aemia a 18mon hs, adjus ed o o he p edic o s (i.e. 1 SD highe WLZ a 6mon hs was associa ed wi h 20% dec ease in p e alence o mala ia pa asi aemia a 18mon hs) (Tables3, 4). Associa ion o WAZ a 6mon hs wi hmala ia om6 o18mon hs In bi a ia e analysis, a highe WAZ a 6mon hs was asso- cia ed wi h lowe incidence o con i med mala ia om 6 o 18mon hs (Table2, unadjus ed), and lowe p e alence o mala ia pa asi aemia a 18mon hs (Tables3, unad- jus ed, 4), bu no incidence o ‘p esumed’ mala ia no incidence o clinical mala ia (Table2, unadjus ed). When adjus ed o o he p edic o s, hese associa ions we e no longe e iden (Tables2, 3, adjus ed and 4). Independen p edic o s o mala ia Independen p edic o s ha we e signi ican ly associ- a ed wi h mala ia ou comes a e p esen ed in Tables3, 4. Incidence o ‘p esumed’ mala ia om 6 o 18mon hs was independen ly associa ed wi h ma e nal age and asse sco e a 6mon hs (i.e. each yea highe in ma e - nal age was associa ed wi h 1% inc ease in incidence o ‘p esumed’ mala ia; and 1 SD highe asse sco e was associa ed wi h 14% dec ease in incidence o ‘p esumed’ mala ia). Incidence o clinical mala ia om 6 o 18mon hs was independen ly associa ed wi h Hb concen a ion and HFIAS a 6mon hs (i.e. each g/L highe Hb a 6mon hs Fig. 1 Flow cha o he child en en olled and included in he inal analysis. The igu e shows he numbe o child en en olled, child en los o ollow up, and child en who we e e en ually included in he analysis om he iLiNS DOSE and iLiNS DYAD-M coho s Page 6 o 12 Bendabendae al. Mala J (2019) 18:143 was associa ed wi h 1% dec ease in incidence o clinical mala ia; and 1-uni highe HFIAS was associa ed wi h 2% inc ease in incidence o clinical mala ia). Highe HFIAS ep esen s highe deg ee o ood insecu i y. The e was also a signi ican a ia ion in he incidence o clinical mala ia be ween he heal h acili ies (s udy si es), wi h Namwe a si e epo ing highe incidence o clinical mala ia compa ed o Mangochi, while he o he wo si es (Malindi and Lungwena) had lowe incidence o clinical mala ia compa ed o Mangochi. Incidence o con i med mala ia om 6 o 18mon hs was independen ly associa ed wi h i on de iciency, ma e nal educa ion and asse sco e a 6mon hs (i.e. i on de iciency a 6mon hs was associa ed wi h 20% highe incidence o con i med mala ia; each yea highe in ma e nal schooling was associa ed wi h 3% dec ease in incidence o con i med mala ia; 1 SD highe asse sco e was associa ed wi h 27% dec ease in incidence o con i med mala ia). The e was also a signi ican a ia- ion in he incidence o con i med mala ia be ween he s udy si es wi h Namwe a si e epo ing highe inci- dence o con i med mala ia compa ed o Mangochi, while he o he wo si es (Malindi and Lungwena) epo ed lowe incidence o con i med mala ia com- pa ed o Mangochi. Independen p edic o s o mala ia pa asi aemia p e - alence we e WLZ, ma e nal age, ma e nal educa ion, Hb concen a ion, and i on de iciency (i.e. 1 SD highe WLZ was associa ed wi h 20% dec ease in p e alence o mala ia pa asi aemia (see Tables3, 4); each yea highe in ma e nal age was associa ed wi h 3% dec ease in p e a- lence o mala ia pa asi aemia; each addi ional yea in ma e nal schooling was associa ed wi h 10% dec ease in p e alence o mala ia pa asi aemia; each g/L highe Hb a 6mon hs was associa ed wi h 2% dec ease in p e a- lence o mala ia pa asi aemia; and i on de iciency a Table 1 Pa icipan cha ac e is ics a age 6mon hs ITN, insec icide- ea ed bed ne s; LNS, lipid-based nu ien supplemen ; RDT, mala ia an igen apid diagnos ic es ; ZPP, zinc p o opo phy in a Measu ed om unwashed enous blood b A week p io o he day o assessmen Mean (SD) o n (%) N Child ac o s Mean (SD) weigh - o -leng h z-sco e 0.3 (1.1) 2561 Mean (SD) weigh - o -age z-sco e − 0.7 (1.2) 2561 Mean (SD) leng h- o -age z-sco e − 1.4 (1.1) 2561 Mean (SD) haemoglobin, g/L 104 (16) 2523 P opo ion o boys 1265 (49.4%) 2561 P opo ion was ed 50 (2.0%) 2561 P opo ion unde weigh 330 (12.9%) 2561 P opo ion s un ed 725 (28.3%) 2561 P opo ion wi h haemoglobin < 105 g/L 1334 (51.5%) 2523 P opo ion wi h ZPP > 70 µmol/mole haema1567 (66.1%) 2371 P opo ion wi h mala ia posi i e by RDT 349 (14.6%) 2389 Ma e nal ac o s Mean (SD) ma e nal age (yea s) 25.8 (6.2) 2265 Mean (SD) ma e nal educa ion, comple ed yea s 4.4 (3.6) 2425 Socio-economic and household ac o s Mean (SD) Household Food Insecu i y Access Sco e 6.1 (5.6) 2128 Mean (SD) asse sco e − 0.02 (0.99) 2179 Mean (SD) numbe o child en aged < 5 yea s in he household 1.6 (0.7) 2149 P opo ion who slep unde ITN dailyb2065 (81.1%) 2546 En i onmen al ac o s P opo ion who ecei ed LNS in e en ion 1804 (70.6%) 2556 S udy si e Mangochi 1667 (65.2%) 2561 Namwe a 445 (17.4%) 2561 Malindi 120 (4.7%) 2561 Lungwena 324 (12.7%) 2561 Page 7 o 12 Bendabendae al. Mala J (2019) 18:143 Table 2 Associa ion o leng h- o -age, weigh - o -leng h, andweigh - o -age z-sco es a age 6mon hs wi hmala ia incidence omage 6mon hs o18mon hs CI: con idence in e al; IRR: incidence a e a io; LAZ: leng h- o -age z-sco e; WAZ: weigh - o -age z-sco e; WLZ: weigh - o -leng h z-sco e a Adjus ed o he ollowing ac o s a age 6mon hs: sex o he child, haemoglobin concen a ion, i on s a us, mon h o bi h, daily use o insec icide- ea ed bed ne s, dis ance o heal h acili y, s udy si e, asse sco es, ma e nal age, ma e nal educa ion, child en < 5yea s, HFIA sco e, and whe he he child ecei ed he s udy in e en ion (LNS) o no b Incidence a e a io, ob ained using nega i e binomial eg ession. The IRR ep esen s he a e o change in incidence o mala ia o each 1-SD highe LAZ, WAZ, o WLZ Incidence o ‘p esumed’ mala ia (N = 2561) Incidence o clinical mala ia (N = 2497) Incidence o con i med mala ia (N = 2497) Unadjus ed Adjus edaUnadjus ed Adjus edaUnadjus ed Adjus eda IRRb (95% CI) p- alue IRRb (95% CI) p- alue IRRb (95% CI) p- alue IRRb (95% CI) p- alue IRRb (95% CI) p- alue IRRb (95% CI) p- alue P edic o a age 6 mon hs LAZ 0.98 (0.94 o 1.02) 0.370 1.03 (0.98 o 1.09) 0.394 0.91 (0.85 o 0.98) 0.011 0.98 (0.89 o 1.07) 0.703 0.92 (0.87 o 0.97) 0.003 1.00 (0.94 o 1.07) 0.882 WLZ 0.96 (0.92 o 1.01) 0.098 0.97 (0.92 o 1.02) 0.190 0.99 (0.95 o 1.04) 0.762 1.03 (0.94 o 1.12) 0.571 0.99 (0.94 o 1.04) 0.623 1.00 (0.94 o 1.06) 0.950 WAZ 0.97 (0.93 o 1.01) 0.094 1.10 (0.50 o 2.42) 0.810 0.95 (0.88 o 1.01) 0.093 1.07 (0.29 o 3.91) 0.916 0.95 (0.90 o 0.99) 0.030 1.46 (0.48 o 4.40) 0.506 Page 8 o 12 Bendabendae al. Mala J (2019) 18:143 6mon hs was associa ed wi h 111% highe p e alence o mala ia pa asi aemia). Discussion The s udy hypo hesis was ha lowe leng h o age z-sco es (LAZ) and lowe weigh o leng h z-sco es (WLZ) a 6 mon hs will be associa ed wi h a highe incidence o mala ia om 6 o 18mon hs and highe p e alence o mala ia pa asi aemia a 18mon hs. In a sample o 2561 Malawian child en, LAZ a 6 mon hs was no associa ed wi h incidence o mala ia om 6 o 18 mon hs, no p e alence o mala ia pa asi aemia a 18mon hs. Highe WLZ a 6mon hs was associa ed wi h lowe p e alence o mala ia pa asi aemia a 18mon hs (20% dec ease in p e alence o mala ia pa asi aemia a 18mon hs o e e y 1 SD highe WLZ a 6mon hs), bu no wi h incidence o mala ia om 6 o 18mon hs. The longi udinal s udy design helped o de e mine he empo ali y o he associa ions because (1) mala ia ou - comes we e ob ained om 6 o 18mon hs a e iden i- ying he p edic o s a 6mon hs, and (2) mala ia is less Table 3 Associa ion o  leng h- o -age, weigh - o -leng h, and weigh - o -age z-sco es a  age 6 mon hs wi h p e alence o mala ia pa asi aemia a age 18mon hs CI: con idence in e al; LAZ: leng h- o -age z-sco e; PR: p e alence a io; WAZ: weigh - o -age z-sco e; WLZ: weigh - o -leng h z-sco e a Posi i e esul om mala ia mic oscopy eadings b Adjus ed o he ollowing ac o s a age 6mon hs: sex o he child, haemoglobin concen a ion, i on s a us, mon h o bi h, daily use o insec icide- ea ed bed ne s, dis ance o heal h acili y, s udy si e, asse sco es, ma e nal age, ma e nal educa ion, child en < 5yea s, HFIA sco e, and whe he he child ecei ed he s udy in e en ion (LNS) o no c P e alence a io, ob ained using a modi ied Poisson eg ession (wi h a obus a iance es ima o ) [29]. The PR ep esen s he a e o change in p e alence o mala ia pa asi aemia o each 1-SD highe LAZ, WAZ, o WLZ P e alence o mala ia pa asi aemia a age 18mon hs (N = 1662)a Unadjus ed Adjus edb PRc (95% CI) p- alue PRc (95% CI) p- alue P edic o a age 6 mon hs LAZ 0.82 (0.72 o 0.93) 0.003 1.11 (0.93 o 1.33) 0.259 WLZ 1.01 (0.88 o 1.16) 0.877 0.80 (0.67 o 0.94) 0.007 WAZ 0.85 (0.75 o 0.96) 0.007 0.88 (0.75 o 1.03) 0.115 Table 4 Independen p edic o s o incidence andp e alence o mala ia inmul i a ia e analysis O he ac o s included in he models bu no signi ican we e: sex o he child, bi h mon h, numbe o child en in household, and daily use o insec icide- ea ed bed ne s CI: con idence in e al; IRR: incidence a e a io; LNS: lipid-based nu ien supplemen s; PR: p e alence a io; ZPP: zinc p o opo phy in a Incidence a e a io, ob ained using nega i e binomial eg ession. The IRR ep esen s he a e o change in incidence o mala ia ( o each 1-uni highe in he con inuous p edic o s o o each g oup compa ed o he e e ence g oup in he ca ego ical p edic o s), adjus ed o o he a iables b P e alence a io, ob ained using a modi ied Poisson eg ession (wi h a obus a iance es ima o ) [29]. The PR ep esen s he a e o change in p e alence o mala ia pa asi aemia ( o each 1-uni highe in he con inuous p edic o s o o each g oup compa ed o he e e ence g oup in he ca ego ical p edic o s), adjus ed o o he a iables c ZPP > 70µmol/mol heme [28] Incidence o ‘p esumed’ mala ia (N = 2561) Incidence o clinical mala ia (N = 2497) Incidence o con i med mala ia (N = 2497) P e alence o mala ia pa asi aemia a age 18mon hs (N = 1662) IRRa (95% CI) p- alue IRRa (95% CI) p- alue IRRa (95% CI) p- alue PRb (95% CI) p- alue Child ac o s a age 6 mon hs Haemoglobin (g/L) 1.00 (0.99 o 1.01) 0.529 0.99 (0.98 o 1.00) 0.016 1.00 (0.99 o 1.01) 0.845 0.98 (0.97 o 0.99) 0.001 I on de iciencyc1.04 (0.91 o 1.18) 0.587 1.03 (0.83 o 1.28) 0.773 1.20 (1.04 o 1.39) 0.012 2.11 (1.24 o 3.57) 0.006 Ma e nal ac o s a en ollmen Ma e nal age 1.01 (1.00 o 1.02) 0.011 0.99 (0.97 o 1.01) 0.149 1.00 (0.98 o 1.01) 0.200 0.97 (0.94 o 0.99) 0.033 Ma e nal educa ion 0.99 (0.97 o 1.01) 0.427 1.00 (0.98 o 1.03) 0.783 0.97 (0.95 o 0.99) 0.024 0.90 (0.85 o 0.96) 0.001 Household ac o s a en ollmen Household Food Insecu i y Access Sco e 1.00 (0.99 o 1.01) 0.809 1.02 (1.01 o 1.04) 0.022 1.01 (0.99 o 1.02) 0.065 1.02 (0.99 o 1.05) 0.184 Asse sco es 0.86 (0.80 o 0.93) < 0.001 0.94 (0.83 o 1.07) 0.343 0.73 (0.64 o 0.83) < 0.001 0.92 (0.63 o 1.34) 0.666 En i onmen al ac o s LNS in e en ion ( s con ol) 1.00 (0.88 o 1.14) 0.983 1.11 (0.89 o 1.37) 0.343 1.02 (0.86 o 1.22) 0.783 1.07 (0.70 o 1.62) 0.765 S udy si e Mangochi Re e ence Re e ence Re e ence Re e ence Namwe a 0.94 (0.81 o 1.09) 0.457 1.49 (1.20 o 1.85) < 0.001 1.73 (1.50 o 1.99) < 0.001 1.49 (0.99 o 2.24) 0.054 Malindi 1.12 (0.88 o 1.42) 0.337 0.45 (0.27 o 0.73) 0.001 0.58 (0.43 o 0.78) < 0.001 0.74 (0.29 o 1.88) 0.531 Lungwena 0.91 (0.75 o 1.09) 0.303 0.24 (0.14 o 0.41) < 0.001 0.33 (0.25 o 0.44) < 0.001 0.65 (0.36 o 1.19) 0.163 Page 9 o 12 Bendabendae al. Mala J (2019) 18:143 common in child en < 6 mon hs due o he p o ec i e e ec o e al haemoglobin and ma e nal an ibodies [30–32]. Ano he s eng h o he s udy was he enhance- men o mo bidi y ecall using a daily pic o ial calenda which minimized ecall bias associa ed wi h home mo - bidi y assessmen s [22, 23]. The s udy had a long ollow up which allowed assessmen o he associa ions in all seasons, and by pooling da a om wo la ge coho s we ended up wi h a la ge sample size wi h adequa e powe o de ec associa ions. The main weakness o he s udy is he change in he exposu e (an h opome ic s a us) du ing he ollow up. The e was signi ican dec ease in mean LAZ, WAZ, and WLZ om 6mon hs o 18mon hs. Compa ed o base- line, mo e child en had de eloped s un ing, unde weigh , and was ing by 18 mon hs (41.5% s 28.3%, 16.5% s 12.9%, and 5% s 2% espec i ely), indica ing ha some child en became nu i ionally wo se o du ing ollow up which may ha e led o misclassi ica ion bias. The e is possibili y o esidual con ounding due o unmeasu ed ac o s associa ed wi h mala ia including HIV in ec ion [33], polypa asi ism [34], i amin A and zinc de iciency [35–38], and gene ic a ia ions including haemoglobinopa hies [39–41]. P esump i e diagnosis o mala ia may also ha e led o o e es ima ing he ou come whe eas he use o RDT- con i med mala ia may ha e led o unde -es ima ing he ou come because (1) 23% o he child en we e ea ed o mala ia a he hospi al wi hou RDT con i ma ion, and (2) child en who had mala ia episodes bu ecei ed home ea men o no ea men and did no p esen o he hospi als we e missed (only 44% o he mo bidi y epi- sodes we e ea ed a he hospi al). Howe e , he consis - ency o he esul s ac oss he di e en mala ia de ini ions sugges s ha his weakness did no bias he esul s, and conside ing he s eng hs o his s udy, he conclusions a e easonably alid. The inding o no associa ion be ween s un edness a 6mon hs and mala ia in he subsequen 12-mon h pe iod is simila o esul s o p e ious s udies [42–45], which sugges s ha s un ing may be o li le signi icance in mala ia epidemiology. These esul s a e howe e di - e en om hose o o he s udies ha epo ed an inc eased isk o mala ia associa ed wi h s un ing. In he Gambia, a p ospec i e s udy o child en aged < 5yea s epo ed an inc eased isk o mala ia associa ed wi h s un ing (c ude RR = 1.35; 95% CI 1.08–1.69) [9]. The ollow up was e y sho (du ing he mala ia season ha las ed only 20weeks), and al hough he ou comes we e adjus ed o age, sex, and e hnici y, c ude RR we e epo ed. The au ho s also did no adjus o socio-eco- nomic and household ac o s he e o e could no ule ou he in luence o hese con ounde s in hei esul s. In Kenya, s un ing in child en aged 0–36mon hs was asso- cia ed wi h an inc eased odds o mala ia pa asi aemia (odds a io = 1.98, p < 0.0001) [10]. The c oss-sec ional design o his s udy sugges s he obse ed associa ion may ha e been due o e e se causa ion, i.e. he cumu- la i e dele e ious e ec s o mala ia on linea g ow h. In Uganda, mild s un ing (IRR = 1.24, 95% CI 1.06–1.46) and mode a e-se e e s un ing (IRR = 1.24, 95% CI 1.03– 1.48) in child en aged < 2.5 yea s we e associa ed wi h inc eased incidence o mala ia pa asi aemia [11]. How- e e , al hough his was a coho s udy, i was no clea whe he exposu e assessmen p eceded he ou comes, he e o e, he empo al ela ionship be ween s un ing and isk o mala ia was di icul o asce ain. Howe e , a s udy in Papua New Guinea epo ed ha lowe LAZ was associa ed wi h lowe incidence o mala ia in he subsequen one-yea pe iod, a ibu ed o inc eased in e e on γ (IFN-γ) esponse o speci ic mala ial an igens obse ed in s un ed child en, al hough he mechanisms we e no ully explained [46]. Mala ia a ies wi h age (incidence o mala ia dec eases whe eas p e alence o mala ia pa asi aemia inc eases wi h age) [13, 47, 48], he e o e, he wide age ange o he s udy pa icipan s ( om 10mon hs o 10yea s) makes o di i- cul compa isons be ween he s udies. In he Papua New Guinea s udy, he au ho s no ed ha in e e on-γ elease inc eased wi h age, he p e alence a es o splenomegaly and pa asi aemia inc eased wi h age, whe eas he inci- dence o mala ia dec eased wi h age, u he muddling he in e p e a ion o he indings. In his popula ion, supplemen a ion wi h LNS did no al e mala ia an ibody acquisi ion [49], which was a ibu ed o he inding ha he nu i ional supple- men s also did no p omo e in an g ow h [16]. Fu he analysis showed no di e ence in mala ia an ibody acqui- si ion be ween he s un ed and non-s un ed child en a 6mon hs (da a no shown), suppo ing he null indings. In his s udy, highe WLZ a 6 mon hs was associ- a ed wi h lowe p e alence o mala ia pa asi aemia a 18mon hs, when adjus ed o o he p edic o s. Was ing is associa ed wi h low le els o lep in h ough deple ion o a mass [50]. Low le els o lep in can lead o educed immune esponse [51], esul ing in inc eased isk o mala ia associa ed wi h was ing. So a , he e idence o associa ion (inc eased o dec eased isk) o mala ia wi h WLZ has mainly come om c oss-sec ional s ud- ies [10, 52, 53]. A ecen sys ema ic e iew de e mined ha he ela ionship be ween was ing and isk o mala ia is hi he o inconclusi e wi h mos longi udinal s ud- ies epo ing no associa ion [54]. This associa ion was no conclusi e and may be due o chance (e.g., because o mul iple es ing [55]), conside ing ha WLZ was