Specific Within-Domain Cognitive Impairments Predict Depression Severity Six-Months After Stroke

Foll o w ing str oke, chr oni c cognitive impair m e n ts acro ss multiple domains have been ass o ciated with depr es sion. Cur rently, it is unknown if s pe cific s ub t yp es of cognitive impairm en ts differe nt ially r elat e t o p ost-str oke depres s ion severity. This st udy aimed to explore the d i fferential ass o c iations between within-domain cognitive imp airment to depr es sion s everit y six-mont hs aft er stroke.

Par ticipant s ( n = 385, Age M = 73.86 year s [ SD = 12.51], Na t ional Ins t itutes of Healt h S troke Severity M = 6.83 [SD = 6.01]) were recr uited f ro m an ac ut e s t roke ward. Part ic ipant s completed a self-r ep ort m ood meas u re (H os pit a l A nx ie t y and D epression Scale; HA DS) and a stroke-specific co gnit i ve as s e s s men t (Ox f ord Cognit i ve Screen; OC S). Separate multipl e r egress ions predicting depr e ssion were conducted a cros s 1) OC S d omain-specific cognitive impairmen t s of language, memor y, attent i o n, praxis , num eracy a nd executive function , and 2 ) the novel s ub task-sp ecific im pai r ments within each O C S d om ai n . A nxi ety sev er ity and yea r s of educ at ion attain ed were included as c ovar i at es .

Wi t hin-domain impa ir men ts that wer e uniqu ely a s so c iat ed w it h d epress ion severity were calc ulat ion ( b (.57) = 1.44, 95% CI [0.31, 2.5 6], p = .012), episodic m emo ry ( b (.52) = 1.36, 95% CI [0.34 , 2.3 7], p = .009), pictu re naming ( b (.45) = 1.18, 95% CI [0.3 1, 2.0 6], p = .008), number writ ing ( b (.46) = 2.54, 95% CI [ 0.26, 2 .07] , p = .012) , an d v i suospatia l at tent ion ( b (.35) = 1.24, 9 5% CI [0.5 4, 1.9 3], p = .001). A nalysis in poo l ed m ultip ly imputed da t a ( N = 430) corr oborated comp let e c ase anal ysis f indings .

Spe c ific within-do main c ognit iv e impa ir men ts have differ ent ia l r elat ionships with post-str oke depres s ive symp tomology. Ac commodat i n g f or these imp airm e n ts in post- stroke depr ession inter v entions m ay pot entia lly enhance t herapeut ic out c o mes . . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint 3 Keywor ds: str oke , c o gnit i o n, c ognit iv e dy sfunction, post- s tr oke depres s ion , po s t -s t roke anxiety . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint 4

Str ok e is a le ad i n g cau se of dis ability 1 which fr equently results in phy s ic al impairm en ts, s u ch a s hem i p a r es is , a s well as co gnitive impai rment s , including poor memory and att enti on. 2 C ognitive im pai rment is part ic ular l y commo n, w it h prevale nce r ates rangin g fr om 34–60% long term after stro k e. 3,4 One in three str ok e survivors will exper ie n c e depr es sion following stroke, 5 chara cterised by a marked de crease i n mo od and/ or a loss of inter e st in da ily activi t ies . 6 Post- str ok e d e p res s ion has been s ho wn t o negatively impact neur orehabi lit ation and reduce long-t erm gains . 7,8 Crucially , p os t -s t ro ke depres si o n has been r epeatedly asso c iat ed w it h general cognitive impair ment in s yste matic reviews . 9,10 G r eater sever ity of cognitive impairm en t still cor rela tes wi t h lower quality of li fe even b eyond 2-years p ost-str oke, 4 su gg e sting a persi stent moo d-cogniti on link ov er t ime. Research inves tigat i n g t he mood-cog nition link has l a r gel y been examined using s c r eening meas ur es of d omain-general cognition su c h as the M ontr ea l Cog nit iv e Ass e s sment (e.g., Ver meer et al. 11 ). Howev er , pos t -stroke co gnitive impair ment c an occur within single o r mult iple ov erlapping bu t dis t inc t areas of c ognit ion (e.g., memory, att enti on). These ca n b e conceptu a lis ed as ‘dom ai n -s pe cific’ impairment s that may have differential impacts on pos t -stroke fun c t ion. 12,13 O f t he do main-specific r elationships t hat have been inves tigat e d in neurological po pulati ons, unique links have been evidenc ed between depr es sion to verb a l me m or y, wor king memor y, a n d ex e cutive fu nction . 14 M ore recently, investigations i n to acro ss-do main c ognit iv e impai r ments have d e l ineat ed a more granular m ood-cognition r e la t ionship i n str oke s ur vi vor s . Willi am s & Demeyer e 15 foun d that memor y, langu a ge, spatial attent ion, exec ut iv e fu nc t ion, numeracy, and pr axis im pai r ments all predicted increases in depr ession s ever ity, ev en aft er covarying for co-occurr ing a n x iet y. . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint 5 Impor tantly, th e asso c iation b etween these domains and depres s ion had diff eren tial strengt hs (β r ange = 1 .18 [memor y ] – 2.0 6 [ praxis ] ). N ot ably, t here was no significant asso ciation between cognitive i mpair m ent and anxiety s everit y six-month s aft er s t roke after covaryin g for co-occurr i n g d e p res s ion sev er ity. 15 Th i s su ggest s c o g n i t ive impai r ment uniquely relates to post- s t roke depre s si o n but not anxiety sev er ity in this population, and that the s t rength o f this associ at ion may v ar y dep ending on the domain o f cognit i o n impaired. Ho w ever , d omain -spec if i c impairm en ts are composed of s ever al with i n -do main “s ubt ype s ” . For example, post-stro k e language impairmen ts could oc cur pur ely in t erms of difficulti es with se m antic understanding, or as difficulties wi t h s p e e ch pr oduction. To our knowledge, there ar e no inv estigatio n s of t he relat ionship of within - domai n cognitive impairm en ts to post-stro k e depr es sio n. Str oke sur v ivor s experiencing an i mpairment i n o ne bro a d d omain o f cognition, suc h a s m emor y , m ay face challe n ges for v ar io us r eas on s , eac h with distinct implications for th e i r ca r e and functioning. It has been shown that within - domain cognitive impa irments can o c cur independent of one another. 16 M ore pr eci se infor ma tion on which s p ec ific within-do main i m pa irmen t s most str ongly l ink with depr es sion c o uld y ield impor tan t inf o rmat ion on how post-stro k e depr ession is m a int ained and in develop ing more t arget ed intervention techniques. Whereas Williams and D emeyere (2021) 15 explo red a popula t ion wi t h rela t ively mild stroke sev er ity (NIHSS <3 = 65. 9% ) , th ere is benefit in examin ing the mood- cognition relation s hip in a mor e m oderat e ly impair ed s t roke s amp l e. P o ssibly , d i f ferent i al relation s hips emerge in mi ld versus m ore mode r ately impai red stroke sa mples , which i s yet unexamined. . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint 6 In this s t ud y, we aimed t o both r eplic at e t he as so ciation between do m a in-s pe cific cognitive impairmen t to p ost-str oke d epress ion a s in W illiams & Dem ey er e 15 in a separate stroke cohor t, and 2) extend th i s inves t ig at ion by ex plor ing how w it hin-do main cognitive impairm en ts c ont ribut e t o post-stro k e depr es sion usin g a m ore cognitively impair ed sample.

Study Design This stud y pr e s ent s a ret rospe ctive an a ly sis of a cro s s-s e ctional observatio n a l within- subje c t s de sign. The S TR O BE guideline was u sed for repor t ing. Data is freel y avai lable at htt ps : / /www.demen tias plat form.u k / and analy s is code at h ttps:// o sf.io/y r t h4/ Sample Eligibi lity and R ec r uitment Procedure Dat a was retr ospec t iv ely anal y s ed fr om the m ul ti-si t e O xfor d Cognitiv e Screening Pro g r amme 13 (N at ional Res ear c h Et hics Comm i t tees REC refer enc e 18/SC /05 50). Elig ibility criter ia wer e: 1) Age > 18 years . 2) Medical diagnos i s of stroke. 3) Capacit y to pr ov ide in f orm ed consent. 4) Abili ty t o c oncent rate for 15 – 20 min utes. Recruit ment occurred in t he O xfor d J ohn R a d c liffe H ospital ac ut e s t roke w ard between 2012 – 2020. Part ic ipant s con senting to th e Oxfor d Cognitive S cr eening p rogramm e a d dit ionally opted t o comple te a s ix-m onth p ost-s tr oke a s sessment of m ood and cognition, for ming the present study dat a . . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint 7 Study Measures Str ok e char acter is t ics, inc lud i n g sex , year s of edu c at i o n, a ge at t ime of s t ro ke, type of stro k e ( i schaemi c vs haemorr hagic) , and lesion hemisphere were ob ta in ed fro m medical recor ds with par ticipant cons en t. Acu te stroke severity wa s a s s es s ed u sing the clinic ian- rat ed N IH SS 17 fr om 0-42 point s where higher s core s indica t e higher stroke sev er ity. Dep res sion and anxiety s ever i t y w er e meas ur ed on the HADS depr es sion (H ADS-D ) and anxiety (HA DS- A) s ubs cales , re spe ctively . 18 Higher s cor e s indicate worse m ood. Scores greater than 7 on ei ther s ub s cale indic at e clinicall y si gnific ant anx iet y /depr ession. Both t he depr es sion (Cro nbac h ' s α = .76) and a nxiety (α = .87) sub s c ales s how good int ernal con s isten cy s ix-mon ths a f ter str ok e . 19 The Ox f ord Cognit i ve Screen (O CS) 13 meas ur es six cognitive domains (with 11 subta sks) to s creen for c ogn i t iv e impa ir ments commonly cau s ed by stroke : language (pictur e naming, s emant ics, s en tence r eading), numer ac y (num ber writing, calculation), praxis (ge s tu re imita tion ), me mor y ( o rientation, s emant ic m e m ory, episodic r ec o gnit ion memor y), vi suospatial atten tion (sus t ained v isual attention, egocentric neg lec t , al locent ric neglect), and exec u tive f unction (tr ai l -making task) . Subtas k respon s e s are binarised indicating impairmen t (1) o r n o i m pa irme nt (0) based on publis h e d cut -off score s from a nor mat ive s amp l e. 13 Impair men t on any of t he subta sks within a domain constitut ed binary domain-specific impair ment. Sample Size Post-h oc Pow er Analysis . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint 8 Gi ven t his stud y wa s a secondar y a n a ly sis of existing data, s am pl e si ze ( N = 430) wa s deter min ed prio r to analys is. Using th e ‘ pwr’ pa cka ge 20 in R, th e comp le t e c ase data ( N = 385) has 9 9% power to det ec t a n ef fe ct s i ze of 0.25 with an alpha of 0.05 as a m i n i m al ef f ect si ze of int e r es t . Ana ly sis Plan All analys es were perf ormed using ‘ Rstudio’ (version 2022.07. 2) and ‘ R ’ (ver s ion 4.2.2: R Cor e Team , 2 022). De s c r iptive s t atis t i cs were carried out ac r oss all s tu d y va riables. Dir ec t c o mparisons on severity of cog nitive impair m ent, mood, and de m ographic variables b e t ween the p resent str oke s ample and th e separat e s t udy samp le analys ed in Wi lliams & D emeyere 15 were conducted to m or e precisely des cribe the sa mple dif fer enc es (see S u pp l e me nt a r y M a te r i a l s ). A series of multip le l inear r eg r es sion analy s es exa m ined the relations hip of 1) the domain-specific cognitive im pairment s, and 2) the subtask- s pecific c o gnit i ve impairments within each domain to p ost- s t roke depressi o n and anx iet y severity. Un s t andardized b- estimates are repor ted wi th 95% confidence inter v als (CIs). The a ssumpt ions of m ultiple linear regr es sion were tested fo r al l models . The high within-subject cova r ian c e bet ween s elf- repo r ted anxi ety and depress ion 21 can bec o me c on flated with th e varian ce in the as s ociation bet ween d epress ion and cognitive impairmen t i n str oke s ur v ivor s 15 , spuriously in cr ea sin g the stren gt h of th is latter relation s hip when measured. Ther efo r e , t he W illiams a n d Demeyere 15 two -m odel app roach was r eplic at ed here: predicting depr e ssion sev er ity with, and without c ovarying for anxi et y sev er ity. The s e mo del s were labelled “Depression O nly M odels” and “ D epres s ion wit h Anxiety Mo dels ”, respectivel y, t hroug hout . Repor t ing on, and directly comparing, bot h types of mod els, aids in un tangl ing t he association between cognitive impa irment and depressi o n . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint 9 sev er ity from t he confounding ef f ec t of anxi et y severity, ult i m ate ly inc r easing confidence in the findings . Fals e dis c o v er y r ate (FDR ) c o rrections were applied in acros s -d omain analy se s t o redu c e Type-I er ror inf lation when pe rfo rm ing mul tiple t es t s . 22 Any s ignific ant do main- spe c ific pred i ct or variables in t he FD R -corr ected regr e s sion models wer e ex plor ed at the subta sk - s pecific lev el using the sa m e a n a ly sis pr o c ed ure. Determ i n ing model covariat es Pairwise Pear s o n c or relations were p er form ed b e t ween dep ress ion and an xiety sev er ity wi t h age s tr oke , ye ar s of ed uc at ion, NIH SS score s, w it h a t wo-taile d independent t- test for dete r min ing an a s so c iat i o n b etween part i cipant sex. Si gnif i cant ly r elat ed v ariabl es (alpha < .05) were included as covariat es . Since the aim of the pres ent s t udy i s to explor e in greater depth an already ident i fied as so ci at ion between doma in- s pecific c o gnit iv e impairm en t and depres sion s everit y 15 th i s data-d riven a p p roach to covariance was chos en to increase the pr edic t iv e accurac y. 23 Mi s singness of Data Mis s ingne s s wa s c las s if i ed as part i cip ants with incomplete OCS data. Analy si s wa s perf orm ed on b oth complete cas e da t a ( n = 3 85 of 430) and on a dataset generat ed by pooling results from fiv e m ul tiply imp uted dat as ets with a maximum of 50 i ter ations to maximis e use of data and det er mine stability of findings. Imput ations wer e c on duc t ed u s ing pred i ctive mean mat c hing (P MM) us i ng the ‘ mi c e ’ R pa ckage. 24 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint 10

Descriptive Statistics Infor matio n about s t roke type and demogr aphic inf ormati o n a n d i s presented in Table 1. [Table 1] Full details of coho rt compar i sons are in the Su pplement a ry Mater ials . In br i ef , the present cohor t had sig n i f ic ant ly gr ea t er stroke severity and a s i gnificantly g reater pro porti o n of do main-specific cognitive impair m ents c om pared to t he s tr o ke cohort repo rted on by Willi am s and D e m eye r e. 15 The assumpt ions of homos c edasti city , linearity, and non-m ul ticolli n ea rity were met for a ll r egres sion models. However, t he assumpt i o n of nor m a li t y o f resi d uals wa s only part ial ly satis f ied, for all m odels due t o the le ft s kewne ss of self-r epor t m o od data clus t ering aroun d z er o (s ee Supplement a ry Mat er ials ). Association of Cognitive Impairment with D epression Depre ss ion Model Co variat es HA DS-D on ly s ignificantly c o rrelated wit h HAD S -A ( r (383) = .63 p < .001), and years of education ( r (380) = -.01 p = .044) . Th erefor e, year s of edu c at i o n a t tained (h er eafte r ‘education’) is incl u ded as a c o v ar iate for all models predicting depr es sion sev er ity. HAD S-A is included as a covariate f or the Depr es s ion wit h Anxiety Models only. Replicatio n - D om ain-Spec if ic Models P redicting Depre ss ion Se verit y Depression Only Models . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint 11 Wi t hout accounting for c o -oc c ur ring anxiety s everit y , t he domain-specific impairm en ts that were signific an t pre dictor v ar iables were language ( b 1 (0.46) = 1.24, 95% CI [0.33, 2.1 4] , t = 2.70, p = .00 7), nume racy (b 1 (0.53) = 1 . 8 6 , 9 5 % CI [0 .82, 2 .89 ], t = 3.52, p < .001), mem ory (b 1 (0.46) = 1 .35, 95% CI [ 0.45, 2.2 6], t = 2.93, p = .004), and sp atial a t t ent i o n (b 1 (0.42) = 1.27, 95% CI [0.44 , 2.1 1], t = 3.00, p = .003). Neither pr axi s nor executive function were signific ant predictor v ar ia b les in their respec t iv e mo dels. Education w as a non - si gn i f ic ant p redictor variable in a ll mo dels. Every Depr ession Only m odel w as s ig nificant ( p r ange = .0 02 [numeracy] – .021 [language]), except for those inc luding pr ax is and exe c ut iv e fu nc t ion. Ex p l ained variance was small for all mod els ( Adjusted R 2 range = .02 [language] – .04 [numerac y], Supplemental Mat erials ). Pooled imputa t ion r esults a ligned with complet e c ase analysi s ( c om plete c ase t - values r a n ge = 1.56 [ e xecutive function ] – 3.52 [numer a cy]; poo l ed imp ut ation t -values range = 1.50 [executive function] – 3. 57 [ numeracy]: Supplementar y M ater ials ). Depression with Anxiety Models. Aft e r acc o unting for c o- occur ring anxiet y severity, dom a in- s pecific impairment s t hat continued t o be as so ci at ed wi t h depr es si o n s everit y wer e language ( b 1 (0.36) = 0.76, 95 % CI [0.05, 1.5 7] , t = 2.11, p = .03 6), nume racy (b 1 (0.42) = 1 . 2 8 , 9 5 % CI [0 .46, 2 .10 ], t = 3.01, p = .002), mem ory (b 1 (0.36) = 0 .91, 95% CI [ 0.20, 1.6 3], t = 2.53, p = .012), spatial a t tent i o n ( b 1 (0.33) = 0.85, 95 % CI [0.19 , 1.50], t = 2.54 , p = .012). Pr ax i s now additionally predi cted dep ress ion sev er ity ( b 1 (0.42) = 0 . 9 4 , 9 5 % CI [0 .89, 3 .48], t = 2.27, p = .024) , but no t exe cut iv e fu nc t ion. Anxiety sev er ity, but not education, was a signif i cant p redictor in al l mod els ( ps < .00 1). Every Depr ession with Anxiety model was s i gnificant ( ps < .001, Adjusted R 2 r ange = .38 [executive function] – .40 [numer ac y]). Pooled imput at ion results aligned w it h c o mplete . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint 12 ca s e analy s i s (complete ca s e t -va lues range = 1.82 [executive function] – 3.01 [ numeracy]; pooled impu t atio n t - v a lues range = 2.00 [executive function] – 3.31 [ numer ac y]: Supplement al Ma teri al ). Exten sion - Subta s k-Specific M odels Predict i n g D epr ession Severity Depression Only Models. Wi t hout accounting for c o -oc c ur ring anxiety s everit y , t he only s ubt ask-spe ci f ic impairm en ts that were signific an t pre dictor v ar iables in mode ls that signific antly predicted depr es sion s everit y wer e c alculation ( p < .001), episodic memor y ( p = .002) , an d v isuospatial att enti on (p < .001: Table 2). Education was a no n- s ignificant pr edictor in al l models. Praxis is a dom ai n comp ris ing of o ne tes t on the OCS, s o no f urther s ub task-spe cific m odels were perf orm ed. Pooled imput at ion results aligned w ith all complet e c ase analyse s ( c om plete case t - values r a n ge = .85 [o bjec t a t tent i o n] – 3.39 [vis uospatial att ention ]; po ole d imputati o n t - values r a n ge = .68 [spa ce attent ion] – 3.37[vis u o s pat i al att ent ion]: S upple m e nt al Mater i al ). [Table 2] Depression with Anxiety Models. Aft e r acc o unting for c o- occur ring anxiet y severity, pictur e naming ( p = .008), number writing ( p = .011) , c alculation ( p = .012), episodic reco gnition ( p = .009), and visuospatial att enti on ( p < .001: Table 3) still pred i cted depr es sion sev er ity. Anxiety s ev erit y, b ut not education, wa s a si gnific ant predictor v ar iable ( ps < .001). A ll D epress ion w it h Anxi ety Mod els were signific ant ( p s < .00 1) . . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint 13 Pooled imput at ion results aligned w ith all complet e c ase analyse s ( c om plete case t - values r a n ge = .20 [o bjec t a t tent i o n] – 3.50 [vis uospatial att ention ]; po ole d i m putat ion t - values r ange = .36 [o bjec t attent i o n] – 3.37 [vis uospatial att ention ]: Supp le ment ary Mat erials ). [Table 3] Summary of Replication Results and N ovel Depression Severity Results The findings of W illiams and Demeyer e ( 2021) 15 s tu dy wer e replica t ed her e in a mor e mode rat ely impai r ed s t roke cohort . The pr oportion o f concurr ent domain- s pecific impairm en ts, or ‘domain-gener al’ i mpairment , significantly predicted depression sev er ity befor e and af ter covarying for c o- oc c ur ring anxie t y s everit y . A dditionall y, in line with Wi lliams & D emeyere (2021 ) 15 neit he r domain-gener a l impai rment nor any domain-specific cognitive impairmen t pred icted anxie ty se ver ity after co-occurr ing for depr ession sev er ity (see S upplement al Mater i al). In our n ov el withi n -do main analys es , only fiv e of the nove l 11 tested OCS s ubt a sk- spe c ific co gnitive i m pai rment s significant ly pr edicted depression s everit y af t er acc o unting for c o- oc cur ring anxie t y s everity and years of education at tained (Figure 1). These ta sks were calc ulat ion, epis odic re cognition, spatial att ention acc ur acy, pi ctur e naming, and number writ ing. Furt her, each s u bta sk-spe cific co gnitive impair m ent as soc iated with depr es sion with differ en tial s t rengths – with performance on the calculatio n task demonstr ating the stro nges t link (Fig ure 2). [Figure 1] [Figure 2] . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint 14

Domain-Specific Impairments This stud y explored t he association between cognitive impai rment s t o dep ress ion sev er ity s ix-mon ths af ter str ok e. Thi s replication and ext ension of the Wil li ams and Dem ey er e 15 study was perfor med in a s t roke s amp le with m ore fr equen t, and severe, cognitive impairmen t s , while detailin g the contr ibution o f wit hin -domain c ognitive impairm en ts to depression s everit y . A s in Will iams and Demeyer e 15 impairments in memor y, langu a ge, num erac y, spatial attent i on, and praxis d omains pr edic t ed depressi o n sev er ity when account ing f or co-oc cu rring anxiety sev er ity. Si m ilar l y , each asso ciation had differential s t rengt hs , su gge s ting t hat the ef f ec t s of po s t- s t roke c o gnitive imp ai r ment on depr es sion s everit y ar e heterogeno us . Exploring t he l ink bet w een p os t -s t ro ke mood and cognition with highly granular co gnitive outcomes aligns with nat ional guidel ines to conceptualis e cognitive impairm en t a s m ultifactorial f or post-stroke r ehabi li t ati on. 25,26 In contr as t t o W illiams a n d Demeyere 15 do main-level exec u tiv e f unction i m pairm ent was not found t o be a sso c iat ed w it h 6 months depression s everit y in t his s ample. It is pos s ible th a t the r e is a lower perceiv able i m pac t of exe cut iv e fu nc t ion on everyday a c t i vities in the context of more severe s t roke. For instance, a s t ud y of older adul ts with neur oc ognit iv e disorder s f ound that executive function imp a i r men t negat i v ely corr el at ed with per fo rmance on complex, multistep activiti es of daily living (AD L s) but not function a l task s like w a shin g and dres si n g on e s e lf. 27 Therefo r e, the ef f ec t s of t his impair ment may decrease i n p erceptual sa lience a s str oke severity increa se s , e speci ally con s ider ing that survivors of more severe strokes are mor e likely to have as sis t an c e with co mp l ex AD L s. 28 Additionally , t he O C S mea sures execu t i ve fun c t ion v ia only one task switchi ng a ss e s s ment . It is po ssible that ot her as p ec t s of exec utive functions note m easured in a simple tra ils tas k . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint 15 coul d potent ially contr ibute to d ep re s s ion aft er stroke. It is worth n oting i n this c on text that the OCS cal culation t ask may have i n adver tent ly indexed an as so ciation bet ween wor k ing memor y i m pair ment and d ep ress ion severity. Finally , t he role of ef f ort in e xecutive tasks should be con s ider ed, gi ven t hat dep ression more gener a lly is a s s o c iated with r educed effort and increased mental fati gue i n dem a n ding c ognit iv e task s . 4,29 The OC S t ask sw it c h i n g asse s s ment us ing circle s and square s ins t ead of let ter s and num bers as a m eans of reducing verbal ef for t for those with aphasi a; ho w ever , this po ssibly alt ers t he effort required for stroke survivors wi t hout aphasia. This opens up avenues f or fut ure r esearc h to investigate a conceiv able r el at ions hip linking effort , post-str oke depres s ion, and d i f feren t s ub- types of exec u tive f unction. Anxiety and education were included as c ovar i at es in th e se models. Despit e the strong a s s ociation of a n x iety with de p ress ion in stro k e , 30 cognitive im pai r ment still had a greater impact of change on depression severity t han a n x iet y severity – indicated by greater b coefficients i n all model s. This su gg es t s t hat post- stro k e co gnitive impai r m ent is mo s t strongly a sso ciated with depr es sion sever i t y. Within-Domain Cognitive Impairments and Depression Severity In addition t o the r ep l ication o f W illia ms and D emeyere 15 are the no v e l associations identifi ed b e tween perf o rmance on s p ec ific within-dom ai n tas k s on the O CS and depres s ion sev er ity. Impairment s in episodic memory, calc ul at ion, number wr iting, pi cture naming, and vis uospatial att ention task s wer e all uniquely asso ciated with increased depress ion severity – even aft er accounting fo r co-occ u rri ng anxiety s everit y. The poten tial mec h a n i sms underlyin g these asso ciations ar e l ikely dynami c an d manifold. In terms of mem or y, inc r eas ed severity of depr es sive s ympt oms has been . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint 16 asso ciated with red uc ed episodic memor y in bot h s t roke 31 and n on -stro k e pop ula t ions . 32 Theref ore, an acquired episodic r ec o g n i t ion impai rment c o uld ex acerbat e a n d maintain exis t i n g s ymptom s of low moo d, i n t u rn wor s ening episodic memory, mut u ally per petuating one anoth er. Impai red episodic mem or y can als o h i n der even bas ic a s pect s of so c ial ising l ike rem e m ber ing what was said in a c onv er s ation , which ha s been documented t o i n c r eas e so c ial isola t ion and frustr ati on in o ther neuro l ogi cal population s . 33 Str ok e survivors identify mem ory diff i culties a s a long-ter m is sue in patient r eport s of unm et n eeds, nega t iv ely impacting quality of li fe. 34,35 M emo ry impairm e nts hav e a tangible impact in dai ly li f e such a s r emembering ho w and when t o ex ecute daily activities and leis ur e ac t iv it ies , and rem embe ri ng having completed them. 36 This enhanced perceptual salience of mem ory impairme n t c o uld generate emotional d is t ress in resp onse to its impact. Wi t h regards to t he number do ma in, mental ar ith me tic requir e s resour c es from working memor y w h i ch i s thought t o mediate th e m a intenance of depressi o n outside of stroke. 37 L ikewise, the observed a s so cia t ion between post-str oke i m pairm ent s in the number writ ing task and depre ss ion s everity may als o be indicative of thi s pot e n tia l connection to wor ki n g m emory. Alt e rnatively, impaired calc ulat i o n a b i l it y m ay hinder engagement in AD Ls like managing financ e s, ther eby reducing autonomo us par tici p a t ion in so c iet y 38 and worsening mood. Simila rly, the nu mber writing t as k involves tr ans cr i b i n g number s f rom speech. S o, t his a s s o ci ation may additionall y ref lect t hat of a red uc ed capa c it y f or activi t ies r equiring abstract pr oc e s si n g f rom verbal command to motor ac t ion and limi t engagement in f unc t ional A DLs. The language domain featur es pr omi nently. H ere we foun d an observed a s s o ciation between depr es sion and impairm en t s in a pictu re naming tas k of c om mon animals and . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint 17 objects. This finding align s wit h the lit erat ure i mplicating aphasi a as a r i s k f actor for post- stroke depr ession symp tomology. 39 Additional ly, ret rieval o f c o mmon obje cts ma y altern a tivel y fu rthe r index an ass o c iat ion bet ween dep ress ion sympto mology a n d difficulties with verbal m emo ry. The curr ent finding that non -neglect- spe cific per form anc e accura c y wa s a sso ci at ed with increased depres s ion s ympto mology align s wi t h the findings fro m Rock and colleag u es' 40 meta -analys i s, indicatin g persi s ten t deficits t o atten t ion in de pressi o n ext ra to the pr es ence of low mood i t s elf. A t te nt ional m echanis m s are composed of various factors, su c h a s effor t, s u s t a ined at tention, a nd su s c ep tibili ty to distraction – all o f which c an be redu c ed fo l lowing stro k e. 41 Consequent l y, acquired imp a ir ments in these a ttent i onal fac t ors following a s t roke could potent ia l ly lessen the degr ee of positive reinfo r c emen t gained fro m activities , r es ult i n g in low mood . Mor eo v er , the exe rtion r equir ed f or da ily a ct i vities post- stroke may contr i b ute to at t ent ional f at i gu e , p otential ly increas in g th e sev erity of depr es siv e symptom s fur ther . Thi s is p a r ticularly relevant to t he O C S visuo s p a t ial subt ask , the br oken hearts task, being b oth effor tful and visuall y complex whic h m a y exa cerbate th e exertion e lement of spatial a ttention. In contr as t t o W illiams a n d Demeyere 15 i mpairments i n t he mo tor i mita tion t as k – the single tas k compr is in g the pr axis domain – only predicted dep r es sion s ev er ity when ac c ou nting for co-occurr ing anxiet y s everity. This incr ease in alpha ( p ) valu es count ers t he decrease expe cted when adding cova riates in mult iple r egr es sion models. 42 Owing to the direct compar is on of results bet ween the Depr es sion Only models and D ep ress ion with Anxiety models, this indic at es pot enti al coll ider bias of t he anx iet y severity variable, li miting inter nal va lidi ty for the p r axis models. 43 One expla n a t ion for t he weak er asso ci at ion between pr axis impairm ent and depr es s ion in a mor e mo der a t e ly af fected s t roke sa m ple . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint 18 could be that it s funct i o nal i m pac t m ay be les s not ic eable if mod era te stoke leads t o a redu c t i o n i n activit ies requir ing phys i cal coor dination. Clinical Implications and Future Research Met hodological ly, th e s cope of these findings is limited t o on ly those c o gnit i ve sub c o mponents as m easur ed i n t he O C S . Thu s , this is neit her an exhaus t i ve nor compreh ens ive lis t o f al l cognitive i m pairm ents that could potential ly pr edict depressi o n sev er ity, s uch a s speed of inf ormati on processing , auditor y at tention, and visual perception difficulti es. 44 R ather, t he novel asso c i ation s identified between s pecific within-dom ai n spe c ific and depr ession s everit y h i ghlight new pat hways for res ear chin g str ategies to augment po s t -s t roke depres si o n treat ments. Impairment s in th es e spec if i c cognitive abili ties may als o have implications for th os e at r is k of su s t ained depress ion severit y and contr ibute to conceptual mod el s of how depr es s ion is maint ained foll owing s t roke. The magnitude of the as sociations be tween cognitive im pai rments and de pr es sion sev er ity wi t hout accounting for anxiet y wer e relative ly small. 45 This wa s bo th corro borated by findings of the analy s is of the poo l ed i m putat ion d a t a a n d by t he f indin g s of W illiams a n d Dem ey er e 15 . Fut u re rep l ications shou ld ext end th ese results and c on s ider c ombining cognitive impairmen t w it h o ther fac t ors known to exa c er bate low mood in stroke – su c h a s fatigue, ps yc h o s o cial is olation , and so cioeconomic deprivation. 10,46

Conceptu a lly, the dir ec t ion of caus ali ty bet w een t hes e ident if ied as so cia t ions is indeter m inable wi thout expe r imen tal evidence o r longitud inal data. Outside of stro k e, depr es sion ha s been shown to indu c e cognitive c h a n ges , like r educed s pee d of . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint 19 pro c es s ing . 44,47 So, one could pot ent i ally ar gue that depr es sion induce s t hes e co gnitive impairm en ts i r respective o f s t roke. Preliminar y evide nce su g ge st s that acute post- s t roke depr es sion outcomes c o uld predict longer- term po s t- s t roke cognitive o utcomes 48 – yet th i s appro a ch is c on founded by the dr as t ic moo d c han ge following acute strok e th a t do no t reliably predict of lon g er -t er m mo od-outcomes. 49 A ddit ionally, it is r eas onable t o a ssume some directional ef fect f rom impairm ent to d ep ress ion becau se the O CS w as de s igned t o identify pr onounced stroke-specific c ognitive impa irm en ts, rather than po pulat i o n- leve l change s in cognition. Statistically , m i xing sel f -repor t measures (depr e ssion and anxi et y severity) with

meas u res ( educa t ion and co g n i t ive imp a i r men t) i n c r eas es the ris k of common meth od v arianc e ( CMV) bia s. 50 This m eans that variance explai n i n g t he out come could be conflated with t he uns t andardised va riance b etween these methods of outc o me c ollection, redu c ing face val idit y. P re s en c e of C M V bias i s indic ated by the increase in variance of depr es sion s everit y each model explained afte r ac count ing f or co-oc curring anxiety sev er ity. Gi ven t hat s elf- reported dep r es sion a nd anxi et y do frequently c o- occur , 21 CM V bias ma y als o b e add i t iv e in t his context, becau se the OC S is not ba sed on sel f -repor t measures. Analysing cont inuous cognit i ve-o utcome data could increas e sen si t ivi t y o f f indings, mitigating some CM V bias . The pr es e nt an a lysi s of binar is ed s cores i n formed t hos e thr es ho l d s of cognition spe cific ally de sig n ed to deter mine functional im pai rm ents. However, a logic al next step is a r eplic at i o n of t his s t ud y wit h more sens it iv e, continu ous co gnitive measures to explore the asso c iation o f s ub task-spe cific co gnitive abili ty – rather than impairm en t – with post-str oke d epres s ion.

. CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint 20 Do main-s p ec ific co gnitive i m pai rmen t s can differ entially infl u enc e t he s everit y of post-str oke d epres s ive sympt oms. S p ecifically, im pai r ments in cal culation , ep i sodic memor y , pictur e nam i n g, n umber writing, a n d non-neglect-specific visuo s pat ial at te ntion t as ks c an all pred i ct p os t -s t roke depres sion sympt omology, ev en af ter contr ol ling for co-occurr ing anxiety s everit y. Id entifying these no vel asso c iation s could help augment existing ther apeut ic interventions in light of t hese i m pairments, pot entia lly enh a n cing ther apeutic outcom e s for those s t roke s ur viv or s mo s t at ris k fo r depressi o n.

The author s t hank a ll par tici p a n ts and m embers of the Oxford Tr ans lat i o n al Neurops y c ho l o gy Group fo r contr i b ut ions to recru i t ment/t esting. Sources of Funding ND (Ad v anced Fellows hip NIHR 30222 4) and AK (D SE Award NIHR305153 ) ar e funded by NIHR. The view s expres se d in t his pub l ication ar e t hose of the auth o r(s) and not n e ces s ar il y those of t he NIH R, N H S or the UK Depar tment of H ealth and Social Care. Disclosures ND i s a developer of the OCS but does not receive remuner at ion fo r its us e. . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint 21

1. Kat an M, Luft A. Global Burden of Str oke. In: Seminar s in N eur ology. Vol 38. ; 2018: 208- 211 . 2. Feigin VL, Brainin M , Nor rving B, et al. W or ld S t roke Organi z at ion (WSO): gl obal stroke f ac t sheet 202 2. Int J Stroke. 2022;17( 1):18-29. 3. El H us seini N , Kat zan IL, Ro s t NS, et al. Cognitive Impair men t A f ter Ischemic and H emorrhagic St roke: A Scientific S t atement F r om the American Heart A s so ciation/American Stro k e A s so ci at ion. Stroke . 2023;5 4(6) :e272-e291. d oi:10.1161/STR. 00000 00 000 0004 30 4. Ku s e c A, Milosevic h E , Will iams OA, et al. L o ng-term ps y chologi c al out c o mes fo l lowing str oke: The O X- CHR O NIC s tu d y . B MC Neurology . 2023; 23(426). h ttps:// doi.or g/ 1 0.118 6/ s 128 8 3-023- 03463-5 5. H ac ket t ML, Pi ck le s K. Part I: Fr equenc y of D epression after Str oke: An Upd ated Sy stematic Review a n d Meta-Analys i s o f O bservation a l Stu dies . Int J S t roke. 201 4;9(8):1 017-1 02 5. d oi:10.11 11/ ij s .12357 6. A mer ican P sychiatr ic As so c iat i o n D. Diagnostic and Sta tistical Man ual of Mental Dis o rd e rs : D S M -5 . Vol 5. American ps ychiatr ic as s o c iation Washin gton, DC ; 2 013. 7. B a r toli F, Di Br i t a C, Crocamo C, Cler ici M, C a r rà G. Ear ly P o s t- s t roke Depr ession and M ort al i t y: Meta-A naly sis a n d Meta-R e gr ession. Front Ps ychiatry . 2018 ;9. Ac cessed Febr uary 1, 20 24. h ttps:// w ww.front iersin.org/journals / ps y c h i at ry/ ar ticle s /10. 3389 / f psyt.2 01 8.0053 0 8. Paoluc ci S, Io s a M, Co i r o P, et al. Post -stroke D epres s ion Incr e a ses D i s ability M ore Than 15% in Is c h emic Str oke S u rviv o rs: A Cas e- Contr ol S t udy. Front Neurol . 2 01 9; 1 0 . A cc es s e d Febr uary 1, 20 24. htt ps : // w ww.fr ontiersin.or g /art ic les / 10.3 389/fneur .201 9.00926 9. H ac ket t ML, Ander s on C S . Pr edictors of Dep res sion aft er Stroke. Str oke. 200 5;36(1 0):2296- 2 301. doi:10. 1161/0 1.S TR . 0000 18362 2.751 35.a4 10. M edeiros G C , Ro y D , Kontos N , B each SR. Po st-str oke d epres s ion: A 20 20 u pdated re v i e w . Gen Hosp Ps ychiatr y . 2020 ;66 : 70-80. doi:10.10 16/j. genho s p psy ch.2 020.06. 011 11. V ermeer J , R ic e D , McInt yre A , Viana R, Macalu s o S, Teas ell R. Cor relates of depr es siv e s ympto ms in individuals at tending ou t patient stroke clinic s. Di sabil Rehab il . 201 7;39(1): 43-4 9. d oi:10.3 109/0963 82 88.201 6.11 40837 12. N y s GMS, van Za n dv o ort M JE, van der Worp HB , et al. Ear ly cognit i ve impa ir ment p redicts lon g - term d e p ress ive sympt oms and quality of life a fter s tr oke. J Ne ur ol S ci . 200 6;24 7(2):149- 15 6. doi:10.10 16/ j.j ns.2 006.0 4.005 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint 22 13. D emeyere N, Riddoch MJ, Slav kov a E D , Bic ker ton WL , H umphr eys G W . Th e O xfor d C ognitive Screen ( O C S): V ali d a t ion of a stroke-spe cific shor t c o g n i t ive s creening tool. Psychol Asse ss . 2 015;2 7(3): 883-894 . doi:10.1037/pa s00000 82 14. R a p oport M J, McCullagh S, Shammi P , Feins t ein A. Cognitive Impairment A s s o ciated With M ajor Depr es sion Following Mild and Moder ate Traum a tic B rain Inju ry. J Neurop sychiat ry Clin N eur osci . 2005 ; 17(1):6 1-65. doi:10.1 176/jnp .17.1 .61 15. Willia m s OA, D emeyere N. A ssoci at ion of depression and anxiet y wit h c og nit i ve imp ai r men t 6 mo nths afte r str oke. Neurology. 2021;96( 15):e196 6-e1974. 16. M urphy D , Cornf ord E, Higgin son A, N orman A, Long R, Noad R. Oxfo rd co g nitive screen: A c r itica l r ev iew and i ndepend ent psychom etric eva luat ion. J Ne ur opsy chol . 202 3;n/ a(n / a). doi:10 .1111 /jnp.12 31 8 17. B rott T, A dams H P, Oli n g er CP , et al. Measurement s of acut e cer e b ral infa rction: a clinical examination scale. St r o k e . 19 8 9;20(7):8 64-87 0. doi:10.11 61/0 1.S TR. 20.7.86 4 18. Zigmond A S, S nait h RP. The ho spital a n xiet y and depr es sion scale. Acta Psy c hiat r Sc and . 198 3;67(6): 361-3 70. 19. J ohnston M, Pollar d B, Hennessey P. Con s tr uct vali d a t ion of the ho s pital a nxiety and d e p ress ion scale with clinica l pop ula t i ons. J Ps y chos om Re s . 2000; 48(6 ):579 -584. d oi:10.1016/S002 2-3 999(00)0 0102-1 20. C hampely S, Ek strom C , Dalga ard P, e t al. pwr: Basic Fun ction s for Power Analy si s. P ublis hed online M arch 17, 20 20. Ac ces s ed Sept ember 3, 2023. ht tps: / / cran.r- p roject.org/web/ p ac kage s /p wr/index.html 21. A nand D, Wilt J, Reve lle W . Within-subject cova r iation between depres sion - and anxiety- r elated af fect. Cogn Emo t. 2017;31(5) : 1055-1061 . doi:10.10 80/02 6999 31.2 016.118 4625 22. Glickman M E, Rao SR , Schult z MR . F al se discovery rat e c ont rol is a r ec om mended alt ernat ive to Bon fe rr oni -type adjustment s in health stud i es. J Clin Epidem iol . 201 4;67(8): 850-8 57. doi:1 0.101 6/ j.j cl inepi.2014.03.01 2 23. D raper NR, Smith H . Ap plied Regres sion An alys is . John Wil ey & Son s ; 19 98 . 24. B uuren S van, G r oothuis-Oud s h oorn K. mice: Mult iv ar ia t e Impu tat ion by C hained Equ ations in R . J S t at Softw . 2011;45: 1 -67. doi:10.186 37/j ss.v04 5.i03 25. NI CE NI f or H and C E. S t roke rehabilit at ion in adul ts. Publis hed on l ine 2013 : 40. 26. National Clinical Guideline for St r oke 2023.pdf . A cce s s ed Ju l y 14, 2 024. h ttps:// w ww. s t roke guideline.org/app/ up l o a d s /20 23/ 0 4/ Na t ional-Clinical -G uideline-f or - St roke- 202 3.pdf?_gl=1*6j49 3p*_up*MQ..*_ga*MTA 0MTM2M TMw N S 4 xN z Iw O Tc x M TQ 4*_ga_ EE3B Z M VLRT*MT c yMD k 3 M TE0 Ny4xL jAuMTc yMD k 3MT E0 N y 4wLjAuM A.. . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint 23 27. C ornelis E, Gorus E, Van Sc h el ver gem N , D e V riendt P . Th e r elationship between bas ic , instr umental, and advanced ac t iv it ies of daily liv ing and executive functioni ng in ger iatric patients with neur oc o g n i t iv e disord ers. Int J G er iatr Psyc hiat ry . 2019;34(6) :88 9- 899 . 28. Sposito G , N er i A L , Yas s ud a M S. Adva nc ed A c t ivi t ies of D aily Living ( A AD L s) and cognitive p e r for manc e in commu ni ty-dwelling el d erly persons : D ata fr om the FIBRA St udy - UNI CAMP . R ev Br a s Geriatr E Geront ol. 2016;19: 07-20. doi:10. 15 90/1809- 982 3.20 16.1 5044 29. A us t in MP, Mit c hell P, Goodwin G M . Cognitive deficits in depr es sion: pos sible imp l ications for functiona l neuro pa th ology. Br J Ps y chiatry . 2001;178( 3):20 0-206. 30. R a f s t en L , D aniel sson A, Sunnerh a gen KS. An xiet y aft er stroke: A sy s t ematic review and me t a - a n a l y s i s . J Rehabil Med . 2018;5 0( 9):769-778. doi:10 .2 340 / 1650 1977- 2384 31. Lim C, Alexander M P. Stroke and episodic memory disorder s . Neurops yc hologia. 200 9;47(1 4):3045- 3 058. doi:10. 1016/j.n europsy c ho l o gi a.2 009.0 8. 002 32. Fang J, D emic S , Chen g S. The reduct i o n of adult n eur ogenes is in depres s ion impair s t he r etrie val of n ew as well as r emote epi sodic memor y . PLOS ONE . 20 1 8;13(6) :e019840 6. d oi:10.1371/ jou rnal.pone.019 8406 33. D avi d s on P, D rouin H , Kwa n D , Mo s co v it c h M , Rosenbaum RS. Memo ry a s S ocial Glue: C l o s e Int e r personal Relationships in Amnesic Patient s . Fron t Ps ychol . 2012 ; 3. Acce ssed Febr uary 1, 20 24. htt ps : // w ww.fr ontiersin.or g /art ic les / 10.3 389/fp s yg .201 2.00531 34. M cKev it t C, Fud ge N, Redf ern J, et al. Self-Rep o rted Long-Ter m N eed s Af te r Str oke. St roke. 2011; 42(5):13 98 -1403. doi:10 .1161/ STR O K EA HA.110. 598839 35. Kim K T, C han g WK, Jun g Y S, et al. U n met Needs for Rehabili tative Management in C ommon H ea l t h-Re lated Pr obl ems Negatively Impact the Quality of Life of Comm unity- D welli n g St roke Survivors. F r ont Neur ol. 2021;12. Ac ce ssed Augu st 23, 20 2 3. h ttps:// w ww.front iersin.org/articles / 10.3389 /fneur.20 21.7 58536 36. Tom aszew ski Farias S , Cahn- Weiner D A, Har v ey D J, et al. L on gi t ud inal Changes in M emory and Executive Func t ioning a re Ass ociated with lon g it udinal c hang e in instr umental activit ies o f daily living i n older Adu l t s . Clin Neuropsyc hol . 20 09;23( 3):44 6- 461 . d oi:10.10 80/ 1 3854 0 4080 23605 5 8 37. Nebe s RD, But ters MA , M uls ant BH, e t al. D ec r eas ed wor ki n g m e m ory and proces s ing speed mediate cognitive impa irm en t in geriatr ic depr es sion. Psyc hol Me d . 200 0;30(3): 679-6 91. doi:1 0.101 7/ S0 0 332917 990 01 968 38. Ko s tan jsek N . U s e of Th e In ternat i o na l C l as si f ic at i o n of Functioning, Disabili t y a n d Health ( ICF) a s a con ceptual framewor k a n d c om mon language for disabili t y s tat ist i cs and h e alt h inf orm a tion s y stems. BMC Pub lic Healt h . 2011;11(4):S3. doi:1 0. 1186 / 1471- 245 8- 11- S4-S3 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint 24 39. Zanella C, Laures- G or e J, Dot s o n VM, Belagaje SR. Inci den c e of po s t - s t roke depressi o n s ympto ms and pot ential r isk fa c t ors in a d ults with ap has ia in a c om prehensive str oke cent er. Top Stroke Rehabil . 2023;30(5) :448-458. doi:10.1 080 /10749 357.2 0 22.2070 36 3 40. R oc k PL, Roiser JP, Riedel WJ, Blackw ell AD . C ognitiv e i m pai rment i n d epr e s s ion: a s y s t ematic review and m eta -analys i s. Ps ychol Med . 2014;4 4( 10):20 29-2040 . d oi:10.1017/S003 3291 7130 0 2535 41. B a r ker-Collo SL, Fei gin V L , Lawes CM M, Parag V, Senior H . Attention Defici ts A fter In c ident S t roke in the Acut e P eriod: F requ e n c y Acro ss T y pe s of At ten t i o n a n d R el at ions h i p s to Pat ient Character is t i cs and Fun ctional O ut comes. Top Str oke R ehabil . 201 0;17(6): 463-4 76. doi:1 0.131 0/ t s r 1 706-463 42. Westfall J, Yarkoni T. Statisticall y Con t roll ing for Confounding Co ns t ructs I s H a r der t han You Th i n k . PLO S ONE . 20 1 6;11(3):e0 1 52719. doi:1 0. 1371 / jour nal.pone.015 2719 43. H olmbe r g M J, Ander s en LW. Collider Bias. JA M A . 2022;327(13): 1 282-1283 . d oi:10.1001/ j ama.2022 .1 820 44. Semkovs ka M , Q u i n l ivan L, O ’ Grad y T , et al. Cognitive function fo l lowing a major d e p ress ive epis od e: a s y s t ematic revi ew and m eta-analys is. La nce t Ps y chi at r y . 201 9;6(10): 851-8 61. doi:1 0.101 6/ S2 2 15- 0366(19)30 291-3 45. C ohen J. Statistical Power An alysis for t he Behavioral Sc iences . 2nd ed. R outledge; 1988. d oi:10.4324/9780 20 377 1587 46. Kh edr EM, A bdelrahman A A , Desoky T, Zaki AF, G amea A. Po s t- s t roke depres si o n: f requency, ris k f acto rs, a n d i m pac t on quality of li fe among 103 stroke patient s — h os pit a l- bas ed stud y. Egypt J Neurol Psy chiatr y N eur o s ur g . 202 0;56(1):6 6. d oi:10.1186/s 41983-02 0- 00199-8 47. Kaiser S, Rot h A, Rentr op M, Friederic h H C, Bender S , Weis br od M. Int ra-individual r eac t ion time variabi lity in s chizophr en i a, depr es sion and bor derline perso nality di s o rd e r . Brain Cogn . 2008;66(1):73- 8 2. doi:10.1016 / j.b a n d c .20 07.0 5 .007 48. Tene O, Shenhar-Tsarf a t y S , Kor czyn AD , et al. Depr essiv e S ympt oms Following Stro k e and Tr ans ient Ischemic Att ack: I s It Ti me for a M o re Intensive Treatme n t A ppr oac h ? R es ults From t he TABASCO Coho rt Study. J Clin Ps yc hiat ry . 2016;7 7(5): 427 1. d oi:10.4088/ J CP. 14m0975 9 49. C a eir o L , Fer ro JM , Sant os C O, Fi gueira ML. Depr es sion in acute s t roke. J Ps yc hiat ry Neuro sci. 20 0 6;31(6): 377- 383. 50. Tehseen S , Ramayah T, S ajilan S. Tes t in g and C ontro l ling for Common M eth od V ariance: A Revi ew of Avai lable M ethods. J Ma nag Sci. 2017;4(2): 142-168. d oi:10.20547 / jm s.2014 .1704 202 . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint 25 Table 1 : Stroke and Dem o gr a p hic In form a t i on Stroke Type n (%) Ischa e mic 255 (6 6.23) Haemorrh agi c 62 (16 .10) Undet ermined from sca n 65 (16 .85) Mixe d 3 (0.7 8) L es i on H em i s phe r e Righ t 152 (3 9.49) Le ft 137 (3 5.58) Bila t e r a l 66 (17 .14) Undet ermined from sca n 30 (7. 79) Stroke R ecurr ence Fi r s t 264 (6 8.37) Rec urrent 121 (3 1.43) Demograp h ics Age ( Y ea r s ) 73.86 (12.51) Educ ation A ttain ed (Ye ar s) 12.25 (3.55) NIHSS S cor e 6.83 (6.01 ) Male Sex (%) 54.03 Note . Thi s tabl e s u mmar ise s c omplet e c a se da ta ( n = 385) , OCS = Oxfo rd Cogni t iv e Sc reen, N IHSS = Natio nal In s t itu te f or Hea lth S trok e Sev e rity . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint 26 Table 2 : ‘Depr es s i o n Only Mod el s’ Predic t i n g Depr es s io n Sev erity by Subta sk-Sp e cific OCS Cog ni tive Imp ai r me n t s an d Educ a tion D ep r e s s i o n Se v er i t y ~ b 0 + b 1 Impai rment + b 2 Educ ation + Di,A Overall Mod el S t a t i s tic s b 1 I m p a ir m ent St at is t i c s Impa irment F st ati s tic ( df ) AdjR 2 p va l u e b 1 (SE ) t s tati stic 95% CI p valu e Pi ctur e Naming 3.7 8 (2, 378 ) .0 1 .260 1 .09 (0.57 ) 1.91 [-0 .03, 2.22 ] .0 57 S emantic Ta sk 4.3 1 (2, 227 ) .0 3 .160 3 .18 (1.25 ) 2.53 [0.70, 5 . 65] .01 2* S ente nce Rea ding 3.3 8 (2, 376 ) .0 1 .385 0 .88 (0.54 ) 1.64 [-0 .17, 1.93 ] .1 01 Number W riting 5.4 1 (2, 373 ) .0 2 .053 1 .58 (0.58 ) 2.71 [0.43, 2 . 73] .00 7* Ca lcula t i on 7.7 3 (2, 372 ) .0 3 .006* 2 .48 (0.72 ) 3.44 [1.06, 3 . 89] < .001* Orientat ion 2.6 7 (2, 377 ) .0 1 .775 0 .71 (0.58 ) 1.23 [-0 .43, 1.86 ] .2 18 Verb al Memo ry 3.0 4 (2, 370 ) .0 1 .538 0 .77 (0.56 ) 1.38 [-0 .33, 1.87 ] .1 68 Epi s odi c Mem ory 6.9 6 (2, 370 ) .0 3 .012* 2 .03 (0.65 ) 3.11 [0.75, 3 . 31] .00 2* Broke n He art Ta sk 8.0 6 (2, 372 ) .0 4 .004* 1 .52 (0.45 ) 3.39 [0.64, 2 . 40] < .001* S pace Attention 2.4 6 (2, 368 ) .0 1 .956 0 .54 (0.64 ) 0.85 [-0 .71, 1.79 ] .3 97 Objec t Attention 2.5 0 (2, 368 ) .0 1 .922 0 .45 (0.50 ) 0.89 [-0 .54, 1.43 ] .3 74 Note . Thi s tabl e s u mmar ise s 11 univa r i at e model s predic ting de pre s sion severi ty w ith 11 o f t h e nov el OCS sub ta s k impa ir me n ts w here th e doma in-l eve l impai r me nts we re s ig ni fic a nt, cov arying fo r educ atio n. ‘ Brok en He art s’ i s a ta s k index ing v isuospa t i a l at ten tion. ‘ S pac e’ a nd ‘Obje ct At t e nti on’ are t ask s ind e xing egoc entr i c and all oc en tr i c ne glec t, r e sp ec tively . *Si gni fic an t (p < .0 50) a ft er f al se discov ery ra te c orrec ti on fo r multipl e compa r i son s. . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint 27 Table 3 : ‘Depr es s i o n with An xiety Model s’ Predic ti ng De pre s s i on Se veri ty b y Sub t a sk-S pec ifi c OCS Cognitiv e Imp ai r me n t, Ed uc atio n, a n d A nxie t y Sev erity Depr e s s io n Sev e r i t y ~ b 0 + b 1 Impa i r m ent + b 2 E ducatio n + b 3 A nxie ty S eve r ity + Di,A Overall Mod el S t a t i s tic s b 1 I m p a ir m ent St at is t i c s Impa irment F st ati s tic ( df ) AdjR 2 p va l u e b 1 (SE ) t s tati stic 95% CI p val ue Pi ctur e Naming 86.6 5 (3, 377 ) .4 0 < .001* 1 .18 (0.45 ) 2.66 [0.31, 2 . 06] .0 08* S emantic Ta sk 48.1 6 (3, 226 ) .3 8 < .001* 1.37(1 .01 ) 1.35 [-0 .63 , 3.36 ] .1 79 S ente nce Rea ding 82.3 9 (3, 375 ) .3 9 < .001* 0 .47 (0.42 ) 1.13 [-0 .35 , 1.30 ] .2 60 Number W riting 85.1 6 (3, 372 ) .4 0 < .001* 1 .17 (0.46 ) 2.54 [0.26, 2 . 07] .0 11* Ca lcula t i on 83.9 1 (3, 371 ) .4 0 < .001* 1 .44 (0.57 ) 2.51 [0.31, 2 . 56] .0 12* Orientat ion 85.2 4 (3, 376 ) .4 0 < .001* 0 .75 (0.45 ) 1.66 [-0 .14 , 1.64 ] .0 98 Verb al Memo ry 80.1 2 (3, 369 ) .3 9 < .001* 0 .56 (0.44 ) 1.28 [-0 .30 , 1.42 ] .2 03 Epi s odi c Mem ory 83.0 2 (3, 369 ) .4 0 < .001* 1 .36 (0.52 ) 2.63 [0.34, 2 . 37] .0 09* Broke n He arts Ta s k 86.0 5 (3, 371 ) .4 1 < .001* 1 .24 (0.35 ) 3.50 [0.54, 1 . 93] <. 001* S pace Attention 80.3 6 (3, 367 ) .3 9 < .001* 0 .75 (0.50 ) 1.51 [-0 .23 , 1.73 ] .1 32 Objec t Attention 79.1 3 (3, 367 ) .4 1 < .001* 0 .14 (0.38 ) 0.20 [-0 .70 , 0.86 ] .7 16 Note . Thi s tabl e s u mmar ise s 11 model s p redic t i ng dep re s s i on seve rity by 11 of t h e novel O CS sub ta s k impai r me nts wh er e domai n- l eve l i mpairmen t s wer e signi ficant, cova r y in g for anxie ty an d educ atio n. ‘ Brok en He art s’ i ndexe s v is uo spa tia l a tte ntion . ‘S pac e ’ and ‘ Objec t A tten t io n ’ index ego centr ic and al locen tric negle c t, r es pe ctive ly. * Signi fica nt ( p < .050 ) a fte r fa l se dis c overy r a t e co r rec t i on fo r multiple comp ari son s. . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint 28 Figure 1 : Raincloud P l ots of Six-Month Post -Stroke Depression Sev erity by Cogniti v e I mpair ment Not e . HADS-D = Hospital Anxiety and Depression Scale-Depression . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint 29 Figure 2 : H eatmap of t he Domain- and Subt ask-Specific Impairment s P redictiv e of D epre ssi on Sev eri ty Six- Mont hs Af ter strok e . CC-BY-NC-ND 4.0 International licenseIt is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.(which was not certified by peer review)preprint The copyright holder for thisthis version posted September 26, 2024. ; https://doi.org/10.1101/2024.09.25.24314204doi: medRxiv preprint