1
Vol.:(0123456789)
Scien i ic Repo s | (2023) 13:7221 | h ps://doi.o g/10.1038/s41598-023-34412-4
www.na u e.com/scien i ic epo s
Using ne wo k analysis o explo e
he alidi y and in luen ial
i ems o he Pa kinson’s Disease
Ques ionnai e‑39
Aline Schönenbe g
1*, Diego San os Ga cía
2, Pablo Mi
3,4,5, Jian‑Jun Wu
6,
Kons an in G. Heim ich
7, Hannah M. Mühlhamme
1,7 & Tino P ell
1
Quali y o li e (QoL) in people wi h Pa kinson´s disease (PD) is commonly measu ed wi h he PD
ques ionnai e‑39 (PDQ‑39), bu i s ac o s uc u e and cons uc alidi y ha e been ques ioned. To
de elop e ec i e in e en ions o imp o e QoL, i is c ucial o unde s and he connec ion be ween
di e en PDQ‑39 i ems and o assess he alidi y o PDQ‑39 subscales. Wi h a new app oach based
on ne wo k analysis using he ex ended Bayesian In o ma ion C i e ion G aphical Leas Absolu e
Sh inkage and Selec ion Ope a o (EBICglasso) ollowed by ac o analysis, we mos ly eplica ed
he o iginal PDQ‑39 subscales in wo samples o PD pa ien s ( o al N = 977). Howe e , model i was
be e when he “igno ed” i em was ca ego ized in o he social suppo ins ead o he communica ion
subscale. In bo h s udy coho s, “dep essi e mood”, “ eeling isola ed”, “ eeling emba assed”, and
“ha ing ouble ge ing a ound in public/needing company when going ou ” we e iden i ied as highly
connec ed a iables. This ne wo k app oach can help o illus a e he ela ionship be ween di e en
symp oms and di ec in e en ional app oaches in a mo e e ec i e manne .
Pa kinson’s disease (PD) is a ch onic p og essi e neu odegene a i e disease cha ac e ized by mo o - and non-
mo o symp oms a ec ing quali y o li e (QoL). Se e al scales ha e been p oposed o assess QoL o people
wi h PD1. Among he exis ing measu emen s, he 39-i em Pa kinson’s Disease Ques ionnai e (PDQ-39) is he
mos ho oughly es ed and applied ques ionnai e, and i is ecommended o assess heal h- ela ed QoL in PD2.
The PDQ-39 was de eloped in 1995 by Pe o e al.3,4 and consis s o 39 i ems, di ided in o 8 subscales: mobili y
(MOB, 10 i ems), ac i i ies o daily li ing (ADL, 6 i ems), emo ional well-being (EMO, 6 i ems), s igma (STI,
4 i ems), social suppo (SOC, 3 i ems), cogni ion (COG, 4 i ems), communica ion (COM, 3 i ems), and bod-
ily discom o (BOD, 3 i ems). Fo each i em, answe s a e gi en on a 5-poin Like scale: ne e , occasionally,
some imes, o en, and always. The sco es om each domain a e compu ed in o a sco e anging om 0 (bes ) o
100 (wo s ). In addi ion, a single index was desc ibed as o e all measu e o heal h- ela ed QoL4. The equen ly
ansla ed PDQ-39 is o e all conside ed eliable, alid, comp ehensi e, and sensi i e o change1,2. Howe e ,
se e al impo an psychome ic issues ega ding de ails o he PDQ‐39 ha e no been su icien ly add essed. In
pa icula , s udies add essing he alidi y o he PDQ-39’s dimensionali y ha e shown inconclusi e esul s5–7,wi h
i ems’ g oup assignmen ailing o ma ch he assumed PDQ‐39 subscales, and eigen alues o se e al ac o s only
ma ginally exceeding 17. The PDQ-39’s scale eliabili y was also subop imal o SOC, COG and BOD7. Con-
side ing hese aspec s is impo an , as eliabili y is c ucial in clinical s udies using he PDQ-39 as a measu e. In
addi ion o an o e all sum sco e, i is c i ical o comp ehend wha he subsco es ep esen . This is a ma e o
cons uc alidi y. Fo ins ance, in p e ious s udies, he COG subscale was mo e closely linked o dep ession
han o neu ocogni i e es s8 o wi h sleep dis u bances and hallucina ions9. The e o e, addi ional s udies a e
OPEN
1Depa men o Ge ia ics, Halle Uni e si y Hospi al, Halle (Saale), Ge many. 2Depa men o Neu ology, Complejo
Hospi ala io Uni e si a io de A Co uña (CHUAC), A Co uña, Spain. 3Unidad de T as o nos del Mo imien o,
Se icio de Neu ología y Neu o isiología Clínica, Ins i u o de Biomedicina de Se illa, Hospi al Uni e si a io
Vi gen del Rocío/Consejo Supe io de In es igaciones Cien í icas/Uni e sidad de Se illa, Se ille, Spain. 4Cen o
de In es igación Biomédica en Red Sob e En e medades Neu odegene a i as, Mad id, Spain. 5Depa amen o
de Medicina, Facul ad de Medicina, Uni e sidad de Se illa, Se ille, Spain. 6Depa men o Neu ology and
Na ional Clinical Resea ch Cen e o Aging and Medicine, Huashan Hospi al, Fudan Uni e si y, Shanghai,
China. 7Depa men o Neu ology, Jena Uni e si y Hospi al, Jena, Ge many. *email: aline.schoenenbe g@
uk-halle.de
2
Vol:.(1234567890)
Scien i ic Repo s | (2023) 13:7221 | h ps://doi.o g/10.1038/s41598-023-34412-4
www.na u e.com/scien i ic epo s/
needed o assess he cons uc alidi y o he PDQ-39 o unde s and he i em alloca ion o he subscales, as well
as hei ela ion o each o he .
Ne wo k analysis is a ela i ely new and p omising me hod o modeling in e ac ions be ween la ge num-
be s o a iables. Ins ead o ying o educe he s uc u e o he a iables o hei sha ed in o ma ion (such as a
la en a iable), as is done in abo emen ioned modeling, he ne wo k es ima es he ela ion be ween all a iables
di ec ly10–12. Ne wo k elemen s (e.g., i ems o a sco e o symp oms) a e concep ualized as elemen s o a complex
in e ac ing dynamical sys em11. Ne wo ks consis o nodes ep esen ing obse ed a iables, and he ela ion-
ships ( o mally called edges) be ween hesenodes. Ins ead o isualizing he ela ionships o all a iables wi h
each o he , ne wo ks can be es ima ed using egula iza ion echniques o igina ing om he ield o machine
lea ning. Regula iza ion means ha edges ha a e likely o be spu ious a e emo ed om he model, leading
o ne wo ks ha a e spa se and easie o in e p e 10. Beyond he mo e con en ional me hods, his ne wo k
app oach has he added bene i o mapping ou he connec ions be ween di e en a iables in he ne wo k and
hus helps unde s and which symp oms a e linked. I may also p o ide an unde s anding o how sub-scales o
a ques ionnai e a e connec ed and which i ems a e c oss-domain b idges, hus p o iding iche in o ma ion
abou he ques ionnai e s uc u e10,13.
The pu pose o he cu en in es iga ion was wo- old. Fi s , we applied ne wo k analysis ollowed by con-
i ma o y ac o analyses (CFA) o e alua e he connec ion be ween PDQ-39 i ems and hei subscales. Speci i-
cally, we a emp ed o comp ehend whe he he subsec ions o he PDQ-39 a e embodied in he i em g oupings
in he ne wo k and which i ems ac as links be ween he a ious subsec ions. We especially employed ne wo k
analysis o a ain insigh in he c oss-domain linkages and o de ec e idence o subs i u e i em g oupings. Sec-
ondly, we sough o gain u he unde s anding in o he connec edness o dis inc i e PDQ-39 i ems and hei
pu po ed ole as a ge s o in e en ions o aise quali y o li e in PD.
Resul s
COPPADIS s udy. The ne wo k plo based on da a om he COPPADIS s udy (N = 694 PD pa ien s, see
Supplemen Table1) is displayed in Fig.1 and i s co esponding cen ali y measu es a e shown in Fig.2 ( abu-
la ed in de ail in Supplemen Table2). Visually, he ne wo k can be di ided in o se e al domains ha mos ly i
o he p oposed PDQ-39 subscales by Pe o e al.3. The hickes edges we e ound wi hin he espec i e domains,
al hough he bo de s be ween hese domains we e b idged by a ious c oss-domain associa ions. Fo example,
he i em “isola ed” connec s he i ems in he EMO subscale o he COC and SOC subscales.
When examining s eng h as a cen ali y measu e, high node s eng h indices we e seen o "need going ou ",
"d essing", "dep essed", and "isola ed". A e pe o ming he boo s apped cen ali y di e ence es (Supple-
men Table3), i became e iden ha he mos in luen ial i ems we e no di e en om one ano he in e ms o
hei cen ali y, so no de ini i e o de can be de e mined. Ne e heless, he i ems "need_going_ou ", "d essing",
"dep essed", "isola ed", "walking_mile", "emba assed", "Suppo _ iends", "close_ ela ion", "ge ing_public", and
"d inking_public" we e iden i ied as being he mos connec ed wi hin he ne wo k. Al e a ions in he alue o
hese poin s hus ha e a bea ing on o he nodes wi hin he ne wo k. Con e sely, i ems such as "anxious", "illu-
sions" and "ho _o _cold" (unpleasan ly ho o cold) appea o be less connec ed when aking in o conside a ion
hei s eng h as opposed o o he nodes. The expec ed in luence also e eals ha he i ems " equi e going ou ",
"ge ing d essed" and "dep essed" a e p o oundly in luen ial and well-connec ed nodes wi hin he ne wo k14.
In e ms o alidi y, i is wo h o no e ha acco ding o he ne wo k, some i ems we e connec ed o di -
e en subscales. Fo example, “igno ed” showed connec ions o bo h “communica e” (COM) and “suppo
iends”/“close ela ions” (SOC), as can be seen in he ne wo k plo and he edge weigh s in Supplemen Table3.
This can also be seen by a low i em- o- es co ela ion o = 0.386 o “igno ed” o i s subscale COM (Supple-
men Table4). Also “hold d ink” om he ADL subscale is mo e s ongly connec ed o “d inking public” om
he STI subscale han o he i ems o he ADL scale. “Hallucina ions” a e close ela ed o he “c amp” i em om
BOD han o he i ems o i s COG subscale. Fo an o e iew o de ailed edge s eng hs see Supplemen Table2.
Acco dingly, a CFA e ealed be e model i when “igno ed” was ca ego ized in o he SOC subscale han in o
he COM subscale (Supplemen Tables5 and 6).
Supplemen Figs.1–3 show he esul s o he accu acy and s abili y checks. O e all, he ne wo k model is
s able, many o he iden i ied edges and cen ali y measu es a e signi ican ly di e en om each o he . The case-
d opping boo s ap p ocedu e shows ha CS-Cs o node s eng h and edges we e 0.67 and 0.75, espec i ely.
Chen e al. s udy. A second sample om Chen e al. (N = 283, see Supplemen Table1 o a compa ison
o bo h s udy g oups) was used o con i m he indings o he COPPADIS s udy15. He e, he ne wo k plo (Sup-
plemen Fig.4) was o e all compa able o he COPPADIS s udy, al hough isual di e ences exis be ween he
wo ne wo k plo s. In he second ne wo k, s eng h was again high o “dep essed” as well as “emba assed”,
“communica ion”, “isola ed”, “ge ing_public”, “need_going_ou ”, “ang y”, “anxie y”, “emba assed”, and “d ess-
ing” (Supplemen Fig.5). When looking a cen ali y di e en es s using boo s apping me hods, node s eng h
did no di e signi ican ly be ween he i ems wi h highes s eng h lis ed abo e, sugges ing ha hey all ha e a
la ge amoun o ele an connec ions. Again, he i em “igno ed” showed connec ions o bo h “communica e”
(COM) and “suppo iends” / “close ela ions” (SOC). This was also con i med by CFA whe e he model i
was again be e when “igno ed” was ca ego ized in o SOC subscale han in o he COM subscale (Supplemen
Table7 and 8). Al hough he sample size o he s udy by Chen e al. was lowe han he COPPADIS s udy, he
ne wo k can s ill be ega ded as s able and in e p e able (Supplemen Figs.6–8). The case-d opping boo s ap
p ocedu e shows ha CS-Cs o node s eng h and edge we e 0.52 and 0.52, espec i ely.
3
Vol.:(0123456789)
Scien i ic Repo s | (2023) 13:7221 | h ps://doi.o g/10.1038/s41598-023-34412-4
www.na u e.com/scien i ic epo s/
Discussion
As a as we a e awa e, his is he i s explo a ion o u ilize ne wo k analysis on he PDQ-39. Ne wo k analysis
can shed ligh on he heo e ical s uc u e o QoL and can expose po en ial a ge s o enhancemen in clini-
cal in e en ion s udies16. Ou s udy has wo signi ican esul s in e ms o ou esea ch goals. Fi s ly, we used
mul iple me hodological echniques o display ha he i em "igno ed" has a close ela ion o he SOC subscale
a he han o COM. Secondly, we iden i ied ne wo k hubs ha b ing o ligh symp oms ha ha e a s ong con-
nec ion wi hin he ne wo k and can likely ha e an impac on he QoL in PD All in all, al hough bo h CFA and
ne wo k analysis ende simila esul s in e ms o he composi ion o subscales, hey bo h b ing o h bene icial
knowledge o he examina ion o da a. Ne wo k analysis e alua es he in e connec ion be ween i ems ins ead
o de e mining hei pa icula impac on a sha ed la en a iable. This inco po a es mapping he mo emen o
da a ia, o ins ance, a hi d i em and u nishes ex a isual and s a is ical de ails on how i ems and subscales o
Figu e1. Ne wo k plo . Abb e ia ions: mobili y (MOB), ac i i ies o daily li ing (ADL), emo ional well-being
(EMO), s igma (STI), social suppo (SOC), cogni ion (COG), communica ion (COM), and bodily discom o
(BOD). Leisu e (Leisu e ac i i ies), Looking (Looking a e home), Ca y (Ca y shopping bags), Walking_mile
(Walking hal a mile), Walking_100 (Walking 100 ya ds), Ge ing_house (Ge ing a ound he house), Ge ing_
public (Ge ing a ound in public), Need_going_ou (Need company when going ou ), Wo y_ alling (Wo y
alling in public), Con ined_house (Con ined o he house), Washing (Washing), D essing (D essing), bu ons
(Do bu ons o shoe laces), W i ing (W i ing clea ly), Cu ing (Cu ing ood), Holdd ink (Hold a d ink wi hou
spilling), Dep essed (Dep essed), Isola ed (Isola ed and lonely), Weepy (Weepy o ea ul), Ang y (Ang y
o bi e ), Anxious (Anxious), Wo ied (Wo ied abou he u u e), Fel _conceal (Fel need o conceal PD),
D inking_public (A oid ea ing/d inking in public), Emba assed (Emba assed due o PD), Wo ied_ eac ions
(Wo ied people’s eac ions), Close ela ion (Close ela ionships), Suppo _pa ne (Suppo om pa ne ),
Suppo _ iends (Suppo om amily o iends), allen_asleep (Unexpec edly allen asleep), Concen a ion
(Concen a ion), Poo _memo y (Poo memo y), hallucina ions (D eams o hallucina ions), Speech (Speech),
communica e (Unable communica e p ope ly), igno ed (Fel igno ed), c amps (Pain ul c amps o spasms), Pain
(Pain in join s o body), ho _o _cold (Unpleasan ly ho o cold).
4
Vol:.(1234567890)
Scien i ic Repo s | (2023) 13:7221 | h ps://doi.o g/10.1038/s41598-023-34412-4
www.na u e.com/scien i ic epo s/
a su ey a e in e wined. Consequen ly, CFA and ne wo k analysis can ei he be in ag eemen o no . The e o e,
ne wo k analysis may help di ec in e en ions by mapping he low o associa ion be ween he included i ems12,16.
In bo h s udy coho s, dep essi e mood, eeling isola ed, eeling emba assed, and ha ing ouble ge ing a ound
in public/needing company when going ou we e iden i ied as well-connec ed a iables. This app oach can help
iden i y impo an aspec s o he illness and di ec in e en ional app oaches in a mo e e ec i e manne .
The PDQ-39 assesses di e en symp oms and unc ions wi h a Like -scaled a ing. I is wo h o no e ha
he symp om cen ali y and he mean alue o he symp om a e di e en aspec s and cons uc s which a e only
weakly associa ed17. Fo example, he alue o a PDQ-39 i em can change wi hou ele an changes wi hin he
ne wo k, because he ne wo k po ays he connec ions be ween i ems and he in luence hey ha e on each o he
in e ms o co-occu ence. Thus, di e en conclusions abou single PDQ-i ems o subscales can be d awn when
looking a cen ali y measu es o symp om se e i y o PDQ-39 i ems18. Fo example, “dep essi e” and “need
company when going ou ” bo h ha e high s eng h in he ne wo k. Howe e , mean alue o “need company when
going ou ” was lowe han o “dep essi e” (Supplemen Table1). I ems in he PDQ-39 ha we e de e mined as
being cen al in he ne wo k a e well-connec ed i ems ha a e linked o many o he symp oms o he disease,
hus hey may be conside ed ele an o he whole concep o QoL, bu see also he limi a ions sec ion o he
in e p e a ion o ne wo k analysis. Consequen ly, aiming a hese i ems speci ically could po en ially esul in
mo e success ul and e icien in e en ions, e en hough i is only possible o con i m he causal in e ence wi h
clinical ials and long- e m da a16,19. Gi en he es ic ed esou ces in medical ca e and he need o e icien
he apies, indings om ne wo k analysis can suppo he de elopmen o e ec i e and ailo ed in e en ions, as
i isually and s a is ically displays which i ems a e linked; his p o ides esea che s wi h a help ul s a ing poin
o de elop ailo ed hypo heses19. Ou indings om he COPPADIS s udy da a poin s o ou a eas ha should
Figu e2. Cen ali y plo . Abb e ia ions: Leisu e (Leisu e ac i i ies), Looking (Looking a e home), Ca y
(Ca y shopping bags), Walking_mile (Walking hal a mile), Walking_100 (Walking 100 ya ds), Ge ing_house
(Ge ing a ound he house), Ge ing_public (Ge ing a ound in public), Need_going_ou (Need company
when going ou ), Wo y_ alling (Wo y alling in public), Con ined_house (Con ined o he house), Washing
(Washing), D essing (D essing), bu ons (Do bu ons o shoe laces), W i ing (W i ing clea ly), Cu ing (Cu ing
ood), Holdd ink (Hold a d ink wi hou spilling), Dep essed (Dep essed), Isola ed (Isola ed and lonely),
Weepy (Weepy o ea ul), Ang y (Ang y o bi e ), Anxious (Anxious), Wo ied (Wo ied abou he u u e),
Fel _conceal (Fel need o conceal PD), D inking_public (A oid ea ing/d inking in public), Emba assed
(Emba assed due o PD), Wo ied_ eac ions (Wo ied people’s eac ions), Close ela ion (Close ela ionships),
Suppo _pa ne (Suppo om pa ne ), Suppo _ iends (Suppo om amily o iends), allen_asleep
(Unexpec edly allen asleep), Concen a ion (Concen a ion), Poo _memo y (Poo memo y), hallucina ions
(D eams o hallucina ions), Speech (Speech), communica e (Unable communica e p ope ly), igno ed (Fel
igno ed), c amps (Pain ul c amps o spasms), Pain (Pain in join s o body), ho _o _cold (Unpleasan ly ho o
cold).
5
Vol.:(0123456789)
Scien i ic Repo s | (2023) 13:7221 | h ps://doi.o g/10.1038/s41598-023-34412-4
www.na u e.com/scien i ic epo s/
be he ocus o a en ion; hese include “needing company when going ou ”, “d essing”, “isola ed” and ea ing
dep ession as hese we e s ongly in e linked in he ne wo k. Dep ession in pa icula has p e iously been iden i-
ied as a p edic o o poo e QoL in PD20. Howe e , i is in e es ing ha ocusing on “d essing” (high cen ali y)
a he han on “cu ing” o “w i ing” (low cen ali y) as o he aspec s o daily ac i i ies appea s o ha e a as e
e ec on well-being and o e all QoL in his g oup o PD pa ien s. O cou se, indings om one coho canno be
ully gene alized o o he coho s o PD pa ien s, as a i udes, belie s, and aspec s ega ded as impo an migh
di e om coun y o coun y. This can be seen by sligh ne wo k di e ences be ween he Spanish COPPADIS
coho and he Chinese coho . While “dep essed” can be conside ed highly connec ed in bo h coho s, “com-
munica ion” o example was mo e well-connec ed in he Chinese han in he Spanish coho . Al hough cul u al
aspec s may play a ole o hese di e ences, i is wo h o men ion ha also PD- ela ed cha ac e is ics ma kedly
di e ed be ween bo h coho s. People in he Chinese coho we e younge , had poo e mo o unc ion, mo e
dep essi e symp oms and poo e cogni i e unc ion (Supplemen Table1).
Conce ning he in luence o dep ession on QoL, o u u e s udies, i would be p omising o addi ionally
include measu es o dep ession in he ne wo k analysis in o de o unde s and which dep essi e symp oms a e
closely linked wi h QoL in PD pa ien s. Also he inclusion o o he PD- ela ed ac o s, such as mo o unc ion
o disease s age, o a ne wo k could p o ide use ul insigh o gain be e unde s anding o QoL and ea men
op ions in PD. Fo ins ance, i we know how anxie y o o he non-mo o symp oms a e associa ed wi h QoL in
ne wo ks, we could de elop a ge ed and mo e e icien in e en ions wi h be e ou comes. Howe e , i is wo h
o no e ha a he cu en s a e o i s me hodical de elopmen , ne wo k analysis is jus one echnique beside
o he explo a i e app oaches and needs jus i ica ion by con i ma o y analysis me hods.
Addi ionally, using se e al app oaches we ound ha “igno ed” seems o be mo e closely ela ed o he SOC
subscale a he han he COM subscale. This was demons a ed in he ne wo k analyses and con i med by CFA
as well as C onbach’s Alpha in bo h samples. This is wo h o no e as i in luences he alidi y o he sepa a e
PDQ-39 subscales and suppo s ha e idence ega ding he psychome ic p ope ies o some o he PDQ-39
subscales is ques ionable8. The PDQ-39 was ini ially de eloped om a 65-i em ques ionnai e, which was es ed
in a la ge sample. A ac o analysis om his sample ound 10 ac o s, and 2 ac o s we e emo ed by he au ho s
because hey we e ega ded as epe i i e3. Howe e , se e al s udies ha e shown ha some i ems do no co ela e
well wi h hei espec i e subscales5,7,8. In pa icula , i ems om he COG subscale equen ly co ela e mo e
s ongly wi h o he domains7,8. This was also he case in ou analysis which sugges s ha “hallucina ions” a e
associa ed wi h BOD i ems a he han wi h he o he COG i ems. This means ha he PDQ-39 subscales should
be used wi h cau ion as ou come measu e o heal h- ela ed QoL in PD ials.
Limi a ions. Ou s udy has se e al limi a ions. As explained, cen ali y measu es o espec i e PDQ-39
i ems should be gene alized wi h cau ion, as hey migh depend on he s udied coho (e.g., di e en inclu-
sion and exclusion c i e ia e c.) and hei clinical and cul u al backg ound. Addi ionally, he in e p e a ion o
cen ali y indices o psychological da a is highly deba ed, which is why we op ed agains epo ing all a ailable
indices (such as be weenness o closeness, as hese a e no ecommended o psychological da a)19,21,22. Howe e ,
when assessing s eng h and expec ed in luence, we adhe ed o he s eps desc ibed by Epskamp e al.10 o ensu e
in e p e abili y o he esul s. This includes using boo s apping me hods o assess he s abili y and signi ican
di e ence o indices. Howe e , while cen ali y indices can be use ul o iden i y impo an a eas o an illness o
a psychosocial cons uc o imp o e e ec i eness o in e en ions, hey should be in e p e ed wi h cau ion and
should be aken as an explo a i e app oach. As Dablande and Hinne19 s a e, in ne wo k analysis—especially
in undi ec ed ne wo ks—i is di icul o assess he di ec ion o in o ma ion low h ough he ne wo k. Unless a
ne wo k is con i med wi h longi udinal da a o a clinical ial, i only p o ides in o ma ion abou which i ems
a e linked bu canno di e en ia e whe he i em A in luences i em B o ice- e sa, Dablande and Hinne19 call
his he p oblem o i ems being a “ oo o lea ” node. The e o e, ne wo k analysis is a use ul ool o map ou he
bidi ec ional connec ions be ween symp oms and o gene a e an unde s anding o which i ems and sub-scales
a e ela ed, bu he hypo heses i may gene a e should be ollowed-up wi h con i ma o y me hods. Also e ec s
o di e en language e sions o he PDQ-39 migh cause di e ences in ne wo k pa e ns be ween di e en
coho s. The used samples a e no ully ep esen a i e o he PD popula ion due o inclusion and exclusion c i-
e ia (i.e., age limi , no demen ia, no se e e como bidi ies, e c.). In addi ion, he es ima ion o a s able ne wo k
usually equi es la ge sample sizes and his limi s i s applicabili y o smalle local coho s. Ne wo k analysis
emains an explo a i e app oach, bu —as o all explo a i e me hods—con i ma o y analysis me hods a e a ail-
able (e.g., o es i a ge ing cen al PDQ-39 symp oms uly has a la ge downs eam e ec on o e all QoL).
Conclusion
In wo independen s udy coho s o PD pa ien s, ne wo k analysis iden i ied dep essi e mood, eeling isola ed,
eeling emba assed, and ha ing ouble ge ing a ound in public/needing company when going ou as well-
connec ed i ems o he PDQ-39, meaning ha hese i ems a e closely in e wined wi h o he PDQ-39 i ems. Ou
esul s hus help unde s and how di e en aspec s o QoL a e linked in PD pa ien s. Addi ionally, we con i m
p e ious conce ns ega ding he alidi y o he PDQ-39 subscales, especially he SOC, COM and BOD and COG
subscales. These esul s highligh he need o use he PDQ-39 subscales wi h cau ion as ou come measu e o
heal h- ela ed QoL in s udies wi h PD pa ien s.
Me hods
Samples. We used da a om wo independen samples o PD pa ien s wi hou PD demen ia:
Da a om he COho o Pa ien s wi h PA kinson’s DIsease in Spain (COPPADIS) s udy was used o es ima e
he PDQ-39 ne wo k. In he COPPADIS s udy, PD pa ien s we e ec ui ed om 35 cen e s in Spain be ween

6
Vol:.(1234567890)
Scien i ic Repo s | (2023) 13:7221 | h ps://doi.o g/10.1038/s41598-023-34412-4
www.na u e.com/scien i ic epo s/
Janua y 2016 and No embe 2017 (N = 694 pa ien s)23. The desc ip i e s a is ic o he coho is gi en in Supple-
men Table1. Dis ibu ions o PDQ-39 on he i em and scale le el a e gi en in Supplemen Table4.
Addi ionally, we used da a om a Chinese obse a ional s udy o 283 people wi h PD ec ui ed om he
Depa men o Neu ology o Huashan hospi al a ilia ed wi h Fudan Uni e si y, Shanghai, China, be ween Ma ch
and Augus 201215. The desc ip i e s a is ic o he coho is gi en in Supplemen Table1.
E hics app o al. Da a collec ion o he COPPADIS s udy was app o ed by he espec i e e hics com-
mi ees o each ec ui ing cen e (45 cen e s in Spain, e.g. Hospi ala io Uni e si a io de Fe ol, A Co uña, see
San os-Ga cía e al. (2015)17 o he ull lis ); he s udy by Chen e al. was app o ed by he e hics commi ee o
Huashan Hospi al, Shanghai. Da a collec ion was pe o med acco ding o he Decla a ion o Helsinki and all
included pa icipan s p o ided w i en in o med consen 15,23.
S a is ical analyses. All analyses we e conduc ed using IBM SPSS s a is ics (Ve sion 25), JASP (Ve sion
0.16), and R (Ve sion 4.1.1), especially he R packages boo ne (Ve sion 1.5)10 and qg aph (Ve sion 1.9)24. Fo a
de ailed desc ip ion o ne wo k analysis in R, see Cos an ini e al.25 and Epskamp e al.10. Desc ip i e s a is ics
we e used o ini ially desc ibe he coho s.
As desc ibed abo e, ne wo ks con ain wo undamen al componen s: nodes, meaning he a iables en e ed
in o he ne wo k model, and edges. Edges desc ibe he ela ionship be ween wo nodes. He e, we compu ed an
undi ec ed and weigh ed ne wo k model. The s eng h o he ela ionship be ween nodes is indica ed in e ms
o edge weigh s, wi h blue edges indica ing posi i e ela ions and ed edges indica ing nega i e ela ions. Edge
weigh s (i.e., s eng h o he ela ion be ween nodes) a e e lec ed by he hickness o he edge, wi h hicke lines
indica ing s onge ela ionships24. We used a ne wo k app oach based on he Gaussian G aphical Model as in o-
duced by Lau i zen26, in which each edge ep esen s he pa ial co ela ion coe icien be ween wo nodes while
condi ioning on he o he a iables in he ne wo k10. The nodes a e posi ioned using he F uch e man-Reingold
algo i hm, which o ganizes he ne wo k based on he s eng h o he connec ions be ween nodes. Nodes wi h
s ong simila i y a e posi ioned close oge he . In a simple co ela ion ne wo k, all signi ican and non-signi ican
edges a e shown, causing a jumbled ne wo k wi h much noise. The e o e, ins ead o using simple co ela ions,
we used a egula ized es ima ion me hod called he Ex ended Bayesian In o ma ion C i e ion G aphical Leas
Absolu e Sh inkage and Selec ion Ope a o (EBICglasso)10 o es ima e he pa ial co ela ions be ween all a i-
ables and sh ink he absolu e weigh s o ze o. Hence, edge weigh s which we e sh unken o exac ly ze o do no
ha e o be es ed agains ze o anymo e, alle ia ing he p oblem o mul iple es ing. Consequen ly, he ne wo k
p oduced a model ha is spa se and easie o in e p e , which is especially help ul as he numbe o nodes and
edges in a ne wo k may in luence he skewness o cen ali y measu es19. As he PDQ-39 con ains a high numbe
o i ems, he ex ended Bayesian In o ma ion C i e ia (EBIC) was used as an in o ma ion c i e ion ha akes bo h
model complexi y and model i in o accoun 27. The hype pa ame e was se a 0.528. Missing da a was handled
using he “exclude pai wise me hod”. As Is o anu and Epskamp29 showed in se e al da a simula ions, ans o -
ma ion o no malize o de ed ca ego ical da a did no imp o e ne wo k accu acy, hus we did no pe o m his
s ep and analyzed he da a as is. Thus, as he PDQ-39 con ains o dinal da a, we
In addi ion o isualizing he ela ionships be ween a iables, he ne wo k can also be desc ibed s a is ically
in e ms o edge weigh s and cen ali y measu es. Cen ali y e e s o he ela i e impo ance o a node and i s
connec ion o o he nodes wi hin he ne wo k. O no e, he cen ali y o a node does no equal i s mean alue,
as i does no ep esen symp om se e i y (as pe cei ed by he espec i e s udy pa icipan ) pe se, bu a he
an i em’s connec edness wi h o he i ems in he ne wo k17. A node wi h high cen ali y is he e o e highly con-
nec ed o o he nodes, and changes in his node can be expec ed o quickly sp ead o o he nodes. Nodes wi h
low cen ali y ha e less in luence on he ne wo k. As a deba e exis s on he applicabili y o cen ali y indices o
psychome ic o medical da a19,21,22, we decided o include only s eng h and expec ed in luence as indica o s o
cen ali y. These a e he mos sui able o ou ype o da a, especially when pai ed wi h an EBICglasso app oach
and boo s apping o assess hei s abili y28–30.
The s eng h (also called deg ee in unweigh ed ne wo ks) is he sum o all di ec associa ions a node has wi h
o he nodes. I e lec s he bundled in o ma ion low o a node in he ne wo k, al hough in undi ec ed ne wo ks
i is no clea whe he his s eng h is oo ed in he a iable being a p edic o o o he a iables, o a consequence
o o he i ems19. Clinically, a node wi h a high s eng h has he po en ial o ep esen an impo an ea u e ( o
example as an i em o a psychological cons uc ), because a change in he alue o his node has a s ong di ec
and quick e ec on o he nodes wi hin he ne wo k. We used a boo s apping app oach o iden i y signi ican
di e ences in he s eng h cen ali y measu e o he a iables wi hin each ne wo k10.
To iden i y highly in luen ial nodes mo e eliably, i may be necessa y o dis inguish be ween posi i e and
nega i e edges and he e o e he expec ed in luence cen ali y measu es was de eloped14. The expec ed in luence
he e o e assesses he node´s cumula i e in luence wi hin he ne wo k.
Finally, measu es o ne wo k accu acy and s abili y we e calcula ed. A case-d opping boo s ap p ocedu e was
used o examine he s abili y o node s eng h. This app oach assumes ha a ne wo k is s able i a la ge p opo -
ion o he sample can be excluded om he da ase wi hou obse ing signi ican changes in he indices. Fo a
s able ne wo k, he Co ela ion S abili y Coe icien (CS-C) should lie abo e 0.510. A nonpa ame ic boo s ap
p ocedu e based on 95% con idence in e als (CIs) was used o assess he edge weigh s s abili y, wi h na ow
95% CIs indica ing a mo e us wo hy ne wo k10. Boo s ap s abili y di e ence es was used o es o s a is ical
signi icance bo h be ween edge-weigh s ha we e non-ze o and be ween nodes in e ms o s eng h cen ali y10.
Based on he indings om ne wo k analyses, con i ma o y ac o analyses (CFA) using JASP (Ve sion 0.16)
we e conduc ed wi h a diagonally weigh ed leas squa e (DWLS) es ima o . Goodness-o - i was e alua ed using
he χ2 s a is ic o exac i , compa a i e i index (CFI), Tucke –Lewis index (TLI), and oo mean squa e e o o
7
Vol.:(0123456789)
Scien i ic Repo s | (2023) 13:7221 | h ps://doi.o g/10.1038/s41598-023-34412-4
www.na u e.com/scien i ic epo s/
app oxima ion (RMSEA). TLI and CFI should be > 0.90, and RMSEA < 0.08 o an accep able i . As no uni e sally
ag eed-upon cu -o is de ined o app op ia e ac o loadings—some esea che s sugges a cu -o o 0.431,32, o h-
e s de ine a cu -o o 0.633,34—we did no de ine a cu -o o ac o loadings in ou analysis. Ins ead, we ocus on
o e all model i as desc ibed abo e and in e p e he associa ions be ween i ems and hei espec i e ac o s in
line wi h he con ex and he esul s om ne wo k analysis. As QoL is a highly complex cons uc and he PDQ-
39 con ains a mul i ude o symp oms ha may in luence i , some a ia ion in ac o loadings is o be expec ed.
O no e, he inclusion o adi ional ac o analysis me hods and ne wo k analysis is bene icial as bo h con-
ibu e aluable in o ma ion ha may be conco dan bu none heless di e s in i s in e p e a ion. Unlike ac o
analysis, whe e a da a s uc u e is es ima ed o assess an unde lying la en a iable ha he included da a aims o
explain, ne wo k analysis conside s he ela ion be ween di e en i ems and how hey a e associa ed wi h each
o he , his includes a po en ial display o in o ma ion low om one i em o ano he ia a hi d i em. The eby,
ne wo k analysis can u he mo e map ou he associa ions be ween di e en sub-scales o a ques ionnai e and
iden i y b idging a iables ha connec di e en aspec s o an illness12,16.
Da a a ailabili y
Da a unde lying he analyses a e a ailable om he co esponding au ho s om he espec i e s udies15,23.
Recei ed: 16 May 2022; Accep ed: 29 Ap il 2023
Re e ences
1. Be a di, A. e al. Tools o assess he quali y o li e in pa ien s wi h Pa kinson’s disease: A sys ema ic e iew. Expe Re . Pha ma-
coecon. Ou comes Res. 21, 55–68 (2021).
2. Ma inez-Ma in, P. e al. Heal h- ela ed quali y-o -li e scales in Pa kinson’s disease: C i ique and ecommenda ions. Mo . Diso d.
26, 2371–2380 (2011).
3. Pe o, V., Jenkinson, C., Fi zpa ick, R. & G eenhall, R. The de elopmen and alida ion o a sho measu e o unc ioning and well
being o indi iduals wi h Pa kinson’s disease. Qual. Li e Res. 4, 241–248 (1995).
4. Jenkinson, C., Fi zpa ick, R., Pe o, V., G eenhall, R. & Hyman, N. The Pa kinson’s Disease Ques ionnai e (PDQ-39): De elopmen
and alida ion o a Pa kinson’s disease summa y index sco e. Age Ageing 26, 353–357 (1997).
5. Tsang, K.-L. e al. T ansla ion and alida ion o he s anda d Chinese e sion o PDQ-39: A quali y-o -li e measu e o pa ien s
wi h Pa kinson’s disease. Mo . Diso d. 17, 1036–1040 (2002).
6. Hagell, P., Whalley, D., McKenna, S. P. & Lind all, O. Heal h s a us measu emen in Pa kinson’s disease: Validi y o he PDQ-39
and No ingham Heal h P o ile. Mo . Diso d. 18, 773–783 (2003).
7. Hagell, P. & Nyg en, C. The 39 i em Pa kinson’s disease ques ionnai e (PDQ-39) e isi ed: Implica ions o e idence based medicine.
J. Neu ol. Neu osu g. Psychia y 78, 1191–1198 (2007).
8. Jones, J. D. e al. The cogni ion and emo ional well-being indices o he Pa kinson’s disease ques ionnai e-39: Wha do hey eally
measu e?. Pa kinsonism Rela . Diso d. 20, 1236–1241 (2014).
9. Wal on, C. C., Sze o, J. Y. Y., Shine, J. M. & Lewis, S. J. G. The ‘Cogni ions’ index o he Pa kinson’s Disease Ques ionnai e-39 ela es
o sleep dis u bance and hallucina ions. Pa kinsonism Rela . Diso d. 21, 349–350 (2015).
10. Epskamp, S., Bo sboom, D. & F ied, E. I. Es ima ing psychological ne wo ks and hei accu acy: A u o ial pape . Beha . Res.
Me hods 50, 195–212 (2018).
11. Bo sboom, D. & C ame , A. O. J. Ne wo k analysis: An in eg a i e app oach o he s uc u e o psychopa hology. Annu. Re . Clin.
Psychol. 9, 91–121 (2013).
12. F ied, E. I. S udying men al heal h p oblems as sys ems, no synd omes. Cu . Di . Psychol. Sci. 31(6), 500–508 (2022).
13. Kossakowski, J. J. e al. The applica ion o a ne wo k app oach o Heal h-Rela ed Quali y o Li e (HRQoL): In oducing a new
me hod o assessing HRQoL in heal hy adul s and cance pa ien s. Qual. Li e Res. 25, 781–792 (2016).
14. Robinaugh, D. J., Millne , A. J. & McNally, R. J. Iden i ying highly in luen ial nodes in he complica ed g ie ne wo k. J. Abno m.
Psychol. 125, 747–757 (2016).
15. Chen, K. e al. E alua ion o PDQ-8 and i s ela ionship wi h PDQ-39 in China: A h ee-yea longi udinal s udy. Heal h Qual. Li e
Ou comes 15, 170 (2017).
16. B ingmann, L. F., Lemmens, L. H. J. M., Huibe s, M. J. H., Bo sboom, D. & Tue linckx, F. Re ealing he dynamic ne wo k s uc u e
o he Beck Dep ession In en o y-II. Psychol. Med. 45, 747–757 (2015).
17. Yang, Z., Algesheime , R. & Tessone, C. J. A compa a i e analysis o communi y de ec ion algo i hms on a i icial ne wo ks. Sci.
Rep. 6, 30750 (2016).
18. Gomez, R., S a opoulos, V., Tulle -P ado, D., Schi inski, B. & Chen, W. Ne wo k analyses o in e ne gaming diso de symp oms
and hei links wi h di e en ypes o mo i a ion. BMC Psychia y 22, 76 (2022).
19. Dablande , F. & Hinne, M. Node cen ali y measu es a e a poo subs i u e o causal in e ence. Sci. Rep. 9, 6846 (2019).
20. Opa a, J., B ola, W., Leona di, M. & Błaszczyk, B. Quali y o li e in Pa kinson`s disease. J. Med. Li e 5, 375–381 (2012).
21. Hallquis , M. N., W igh , A. G. C. & Molenaa , P. C. M. P oblems wi h cen ali y measu es in psychopa hology symp om ne wo ks:
Why ne wo k psychome ics canno escape psychome ic heo y. Mul i a . Beha . Res. 56, 199–223 (2021).
22. B ingmann, L. F. e al. Wha do cen ali y measu es measu e in psychological ne wo ks?. J. Abno m. Psychol. 128, 892–903 (2019).
23. San os-Ga cía, D. e al. COPPADIS-2015 (COho o Pa ien s wi h PA kinson’s DIsease in Spain, 2015), a global–clinical e alu-
a ions, se um bioma ke s, gene ic s udies and neu oimaging–p ospec i e, mul icen e , non-in e en ional, long- e m s udy on
Pa kinson’s disease p og ession. BMC Neu ol. 16, 26 (2016).
24. Epskamp, S., C ame , A. O. J., Waldo p, L. J., Schmi mann, V. D. & Bo sboom, D. qg aph: Ne wo k isualiza ions o ela ionships
in psychome ic da a. J. S a . So w. 48, 1–18 (2012).
25. Cos an ini, G. e al. S a e o he aR pe sonali y esea ch: A u o ial on ne wo k analysis o pe sonali y da a in R. J. Res. Pe s. 54,
13–29 (2015).
26. Lau i zen, S. L. G aphical Models (Ox o d Uni e si y P ess, 1996).
27. Foygel, R. & D on, M. Ex ended Bayesian In o ma ion C i e ia o Gaussian G aphical Models. In Ad ances in Neu al In o ma ion
P ocessing Sys ems (eds La e y, J., Williams, C., Shawe-Taylo , J., Zemel, R. & Culo a, A.) ol. 23 (Cu an Associa es, Inc., 2010).
28. Epskamp, S. & F ied, E. I. A u o ial on egula ized pa ial co ela ion ne wo ks. Psychol. Me hods 23, 617–634 (2018).
29. Is o anu, A.-M. & Epskamp, S. Which es ima ion me hod o choose in ne wo k psychome ics? De i ing guidelines o applied
esea che s. Psychol. Me hods h ps:// doi. o g/ 10. 1037/ me 00 00439 (2021).
30. Opsahl, T., Agneessens, F. & Sk o e z, J. Node cen ali y in weigh ed ne wo ks: Gene alizing deg ee and sho es pa hs. Soc. Ne w.
32, 245–251 (2010).
31. S e ens, J. Applied Mul i a ia e S a is ics o he Social Sciences (Law ence E lbaum Associa es, 2002).
8
Vol:.(1234567890)
Scien i ic Repo s | (2023) 13:7221 | h ps://doi.o g/10.1038/s41598-023-34412-4
www.na u e.com/scien i ic epo s/
32. Tabachnick, B. G. & Fidell, L. S. Using Mul i a ia e S a is ics 5 h edn, Vol. xx ii, 980 (Allyn & Bacon/Pea son Educa ion, 2007).
33. Field, A. Disco e ing S a is ics Using SPSS 2nd edn, Vol. xxxi , 779 (Sage Publica ions, Inc, 2005).
34. Guadagnoli, E. & Velice , W. F. Rela ion o sample size o he s abili y o componen pa e ns. Psychol. Bull. 103, 265–275 (1988).
Acknowledgemen
We hank he COPPADIS S udy G oup as well as Jian-Jun Wu and Colleagues o hei ho ough da a acquisi ion
and managemen , and o p o iding us wi h hei da a o his analsysis.
Au ho con ibu ions
T.P. design, execu ion, analysis, w i ing. A.S. analysis, edi ing o inal e sion o he manusc ip . D.S.G. da a
p epa a ion, edi ing o inal e sion o he manusc ip . P.M. da a p epa a ion, edi ing o inal e sion o he
manusc ip . J.J.W. da a p epa a ion, edi ing o inal e sion o he manusc ip . K.H. edi ing o inal e sion o
he manusc ip . H.M.M. analysis, w i ing.
Funding
Open Access unding enabled and o ganized by P ojek DEAL. Funding o K.H. was p o ided by heGe -
man Resea ch Founda ion(DFG) as pa o he Clinician Scien is -P og am O ganAge(413668513). TP, HMM
and KH ecei ed unding oma Bundesminis e ium ü Bildung und Fo schung (BMBF, Fede al Minis y o
Educa ion and Resea ch) g an (01GY1804). T.P. ecei ed unding om a Bundesminis e ium ü Bildung und
Fo schung (BMBF, Fede al Minis y o Educa ion and Resea ch) g an (01GY2301). P.M. ecei ed g an s om
heSpanish Minis y o Science and Inno a ion [RTC2019-007150-1], he Ins i u o de Salud Ca los III-Fondo
Eu opeo de Desa ollo Regional (ISCIII-FEDER) [PI18/01898], he Conseje ía de Economía, Inno ación, Ciencia
y Empleo de la Jun a de Andalucía [CVI-02526, CTS-7685], and he Conseje ía de Salud y Bienes a Social de la
Jun a de Andalucía [PI-0471-2013, PE-0210-2018, PI-0459-2018, PE-0186-2019], J.J.W. ecei ed unding om
heNa ional Heal h Commission o he People’s Republic o China (G an No. P o20211231084249000238), and
o Shanghai Municipal Science and Technology Majo P ojec s (G an No. 2018SHZDZX01, 21S31902200).
Compe ing in e es s
The Au ho s decla e no compe ing non- inancial in e es s bu epo he ollowing inancial disclosu es: AS and
KH epo no disclosu es. DSG ecei ed hono a ia o educa ional p esen a ions and ad ice se ice by Abb ie,
UCB Pha ma, Lundbeck, KRKA, Zambon, Bial, I al a maco, and Te a. PM has ecei ed g an s by he Fundación
Alicia Koplowi zand also ecei ed suppo o a ending mee ings and/o a el o hono a ium o lec u ing
om Abbo , Alle gan, Abb ie. Bial, B i annia, I al a maco, Me z, UCB, Te a and Zambon.
Addi ional in o ma ion
Supplemen a y In o ma ion The online e sion con ains supplemen a y ma e ial a ailable a h ps:// doi. o g/
10. 1038/ s41598- 023- 34412-4.
Co espondence and eques s o ma e ials should be add essed o A.S.
Rep in s and pe missions in o ma ion is a ailable a www.na u e.com/ ep in s.
Publishe ’s no e Sp inge Na u e emains neu al wi h ega d o ju isdic ional claims in published maps and
ins i u ional a ilia ions.
Open Access This a icle is licensed unde a C ea i e Commons A ibu ion 4.0 In e na ional
License, which pe mi s use, sha ing, adap a ion, dis ibu ion and ep oduc ion in any medium o
o ma , as long as you gi e app op ia e c edi o he o iginal au ho (s) and he sou ce, p o ide a link o he
C ea i e Commons licence, and indica e i changes we e made. The images o o he hi d pa y ma e ial in his
a icle a e included in he a icle’s C ea i e Commons licence, unless indica ed o he wise in a c edi line o he
ma e ial. I ma e ial is no included in he a icle’s C ea i e Commons licence and you in ended use is no
pe mi ed by s a u o y egula ion o exceeds he pe mi ed use, you will need o ob ain pe mission di ec ly om
he copy igh holde . To iew a copy o his licence, isi h p:// c ea i eco mmons. o g/ licen ses/ by/4. 0/.
© The Au ho (s) 2023