Efficacy of bio- and neurofeedback for depression: a meta-analysis

Psychological Medicine camb idge.o g/psm Re iew A icle Ci e his a icle: Fe nández-Al a ez J, G assi M, Colombo D, Bo ella C, Cip esso P, Pe na G, Ri a G (2021). E icacy o bio- and neu o eedback o dep ession: a me a- analysis. Psychological Medicine 1–16. h ps:// doi.o g/10.1017/S0033291721004396 Recei ed: 29 No embe 2019 Re ised: 29 Sep embe 2021 Accep ed: 7 Oc obe 2021 Keywo ds: Dep ession; bio eedback; neu o eedback; hea a e a iabili y; hea a e a iabili y bio eedback; MRI neu o eedback; me a-analysis Au ho o co espondence: J. Fe nández-Al a ez, E-mail: ja e [email p o ec ed] © The Au ho (s), 2021. Published by Camb idge Uni e si y P ess. This is an Open Access a icle, dis ibu ed unde he e ms o he C ea i e Commons A ibu ion licence (h p://c ea i ecommons.o g/licenses/by/4.0/), which pe mi s un es ic ed e-use, dis ibu ion and ep oduc ion, p o ided he o iginal a icle is p ope ly ci ed. E icacy o bio- and neu o eedback o dep ession: a me a-analysis J. Fe nández-Al a ez1,2, M. G assi3,4, D. Colombo2, C. Bo ella5, P. Cip esso6,7, G. Pe na3,4,8,9 and G. Ri a1,6 1 Depa men o Psychology, Ca holic Uni e si y o he Sac ed Hea , Milan, I aly; 2 Depa men o Basic Psychology, Clinic and Psychobiology, Uni e si a Jaume I, Cas ellón, Spain; 3 Depa men o Clinical Neu osciences, He manas Hospi ala ias, Villa San Benede o Menni Hospi al, FoRiPsi, Albese con Cassano, Como, I aly; 4 Depa men o Biomedical Sciences, Humani as Uni e si y, Rozzano, Milan, I aly; 5 Cibe Fisiopa ología Obesidad y Nu ición, CB06/03 Ins i u o Salud Ca los III, Mad id, Spain; 6 Applied Technology o Neu o-Psychology Lab, IRCCS Is i u o Auxologico I aliano, Milan, I aly; 7 Depa men o Psychology, Uni e si y o Tu in, Tu in, I aly; 8 Depa men o Psychia y and Beha io al Sciences, Mille School o Medicine, Uni e si y o Miami, Miami, FL, USA and 9 Resea ch Ins i u e o Men al Heal h and Neu oscience and Depa men o Psychia y and Neu opsychology, Facul y o Heal h, Medicine and Li e Sciences, Uni e si y o Maas ich , Maas ich , he Ne he lands Abs ac Backg ound. Fo many yea s, bio eedback and neu o eedback ha e been implemen ed in he ea men o dep ession. Howe e , he e ec i eness o hese echniques on dep essi e symp- oma ology is s ill con o e sial. Hence, we conduc ed a me a-analysis o s udies ex ac ed om PubMed, Scopus, Web o Science and Embase. Me hods. Two di e en s ings we e conside ed o each o he wo objec i es o he s udy: A i s g oup comp ising s udies pa ien s wi h majo dep essi e diso de (MDD) and a second g oup including s udies a ge ing dep essi e symp oma ology educ ion in o he men al o medical condi ions. Resul s. In he i s g oup o s udies including pa ien s wi h MDD, he wi hin-g oup analyses yielded an e ec size o Hedges’g=0.717, while he be ween-g oup analysis an e ec size o Hedges’g=1.050. Mode a o analyses indica e ha ea men e icacy is only signi ican when accoun ing o expe imen al design, in a o o andomized con olled ials (RCTs) in com- pa ison o non RCTs, whe eas he ype o neu o eedback, ial design, yea o publica ion, numbe o sessions, age, sex and quali y o s udy did no in luence ea men e icacy. In he second g oup o s udies, a small bu signi ican e ec be ween g oups was ound (Hedges’g= 0.303) in a o o bio- and neu o eedback agains con ol g oups. Mode a o ana- lyses e ealed ha ea men e icacy was no mode a ed by any o he sociodemog aphic and clinical a iables. Conclusions. Hea a e a iabili y (HRV) bio eedback and neu o eedback a e associa ed wi h a educ ion in sel - epo ed dep ession. Despi e he ac ha he ield has s ill a la ge oom o imp o emen in e ms o esea ch quali y, he esul s p esen ed in his s udy sugges s ha bo h modali ies may become ele an complemen a y s a egies o he ea men o MDD and dep essi e symp oma ology in he coming yea s. Majo dep essi e diso de (MDD) ep esen s a wo ldwide leading cause o disabili y, wi h mo e han 300 million people a ec ed (WHO, 2017). No only does i en ail a majo impac on people’s quali y o li e and social unc ioning (Ange meye , Holzinge , Ma schinge , & S engle -Wenzke, 2002; Gili e al., 2013; IsHak e al., 2013; Zuelke e al., 2018), bu also MDD is s ongly ela ed wi h a as a ay o o he men al diso de s, mainly anxie y diso de s (Wa son, 2009), as well as a g ea numbe o medical condi ions, including ch onic physical illnesses (Kang e al., 2015) and neu ological diseases (Raskind, 2008). Al hough psycho he apy (Cuijpe s, C is ea, Ka yo aki, Reijnde s, & Huibe s, 2016), psy- chopha macology (Cip iani e al., 2018) and he combina ion o he p e ious wo (C aighead & Dunlop, 2014) ha e shown o be e icacious, he e is s ill much oom o imp o emen . A ound 40–50% o pa ien s do no espond o ea men (Cuijpe s e al., 2014) and a hi d o hose who do espond p esen elapses (Beshai, Dobson, Bock ing, & Quigley, 2011; Bu cusa & Iacono, 2007). Besides, i is es ima ed ha 75% o dep essed people emain un ea ed (WHO, 2017). Hence, i is o u mos impo ance o de elop new modali ies o ea men ha can help o o e come he a o emen ioned obs acles (Kazdin & Blase, 2011). In his sense, bio eedback is conside ed one o he exis ing mind-body in e en ions ha may os e he b idging o physiological and psychological in e en ions. Bio eedback echni- ques en ail a signal (e.g. ideo, audio display o ac ile) connec ed o a physiological p ocess ha enables he pe son o be awa e o no mally unconscious physiological ac i i y (B owne, 2015). In his sense, indi iduals a e p o ided wi h explici in o ma ion o a ce ain psycho- physiological p ocess in o de o os e i s egula ion. h ps://www.camb idge.o g/co e/ e ms. h ps://doi.o g/10.1017/S0033291721004396 Downloaded om h ps://www.camb idge.o g/co e. Uni e si a Jaume I, on 12 Jan 2022 a 08:03:41, subjec o he Camb idge Co e e ms o use, a ailable a Bio eedback has p incipally been used in he medical ealm, al hough he e is also a long-s anding adi ion o esea ch on bio eedback echniques o men al diso de s (Leh e & Ge i z, 2014; Sacche & Go lib, 2016). In pa icula , pos - auma ic s ess diso de and subs ance use diso de a e among he mos esea ched condi ions (Schoenbe g & Da id, 2014). Di e en physiological p ocesses ha e been implemen ed o bio eedback p ocedu es, including bo h he cen al and au onomous ne ous sys ems. Elec omyog aphy bio eedback (EMGB), skin conduc ance bio eedback o hea a e a iabili y bio eedback (HRVB) a e some o he mos used pe iphe al esponses, while elec oencephalog aphic (EEG) and unc ional magne ic esonance imaging neu o eedback ( MRI-NF) a e wo o he mos common echniques using neu al ac i i y (Sacche & Go lib, 2016). Ample e idence demons a ed ha di e en psychophysio- logical p ocesses a e impai ed in pa ien s wi h MDD. Wi h ega d o he neu oci cui y, unc ional impai men s ha e been iden i ied in p e on al, limbic, s ia al, halamic and basal o eb ain s uc- u es (P ice & D e e s, 2010). O pa icula impo ance o neu- o eedback, he e is consis en e idence om EEG esea ch demons a ing ha dep essi e indi iduals p esen highe le -hemisphe ic alpha ac i i y, including hypoac i a ion in he le p e on al a ea. In his ega d, an imp o emen o he dep es- si e symp oma ology has been obse ed a e a neu o eedback- based aining o his asymme y (Linden, 2014). Likewise, neu- oimaging and b ain s uc u al esea ch indica e ha people wi h dep ession p esen se e al abno mali ies. Fo ins ance, he amygdala has been iden i ied as an impo an a ge in neu o eed- back in e en ions o dep ession due o i s ole in emo ional p o- cessing and esponding, in e ac ing wi h di e en co ical and subco ical a eas and ha ing shown o be a key ma ke o he onse and eco e y o MDD (Young e al., 2018). Likewise, ca diac ac i i y has p o en o g ea ly con ibu e o he gene al physiological dys egula ion o dep essed pa ien s, and no only o be a co ela e o he neu al dys egula ion (Thaye & Ma he , 2018). Hea a e a iabili y (HRV), in pa - icula he high equency (HF) o he spec al domain, is consid- e ed o index ca diac agal one and hus o be a ele an ma ke o MDD. Resea ch in his domain indica es ha dep ession is associa ed wi h lowe es ing HF-HRV and lowe LF/HF a io (Hamil on & Alloy, 2016; Kemp e al., 2010). Taken oge he , hese esul s indica e ha NF and HRVB con- s i u e wo echniques ha ga he consis en heo e ical suppo o jus i y a psychophysiological in e en ion. Indeed, he e a e a numbe o quali a i e e iews (Hammond, 2005; Linden, 2014; Sacche & Go lib, 2016; Young e al., 2018) ha ha e ga he ed he a ailable s udies o neu o eedback and bio eedback o MDD and dep essi e symp oma ology. Ne e heless, o he bes o ou knowledge, no s udy has me a-analy ically es ablished he ex en o which his app oach is e icacious o MDD and dep essi e symp oma ology, espec i ely. Apa om calcula ing he o e all e ec o bio- and neu o eed- back in e en ions o MDD, his s udy also aims o calcula e he e ec o all bio- and neu o eedback s udies ha included dep es- si e symp oma ology as a seconda y ou come measu e in subjec s su e ing om o he condi ions han MDD. Main esea ch ques ions (1) Wha is he pooled e idence o he e ec i eness o bio- and neu o eedback o MDD? (2) Wha is he pooled e idence o he e ec i eness o bio- and neu o eedback o dep essi e symp oms in bo h medical and men al/psychia ic condi ions o he han MDD? (3) Wha mode a o s explain possible sou ces o he e ogenei y among he e ec sizes? Ma e ials and me hods The sys ema ic e iew has been de eloped in acco dance wi h P e e ed Repo ing I ems o Sys ema ic Re iews and Me a-Analyses (PRISMA) (see Supplemen a y 1) (Mohe e al., 2009). Sea ch s a egy Fi s , a icles we e iden i ied h ough comp ehensi e sea ches o he ollowing da abases: PubMed, Scopus, Web o Science and Embase. The las upda e was in Decembe 2018. Re e ences lis s o e iew a icles we e also conside ed o po en ial unde ec ed s udies and g ay li e a u e has also been examined ( o he sea ch s ing see Appendix 2). Eligibili y c i e ia This s udy ollows a wo-s ep le el s uc u e, and hus wo di e - en eligibili y c i e ia ha e been conside ed. To add ess he i s aim, o iginal a icles in English epo ing da a o he e icacy o bio- and neu o eedback in he ea men o MDD we e consid- e ed. To selec s udies, he e m “clinical dep ession”u ilized in he DSM 5 (Ame ican Psychia ic Associa ion, 2013) was consid- e ed, which comp ises MDD and dys hymic diso de . All s udies ha ei he es ablished a diagnosis o dep ession using a s anda - dized diagnos ic in e iew (such as he SCID, CIDI, o SCAN) o pa icipan s who p esen ed ele a ed symp oms o dep ession based on sel - epo measu es we e conside ed o inclusion. ) S udies ha included subjec s aking psychopha macology o ecei ing any o he ac i e ea men such as ho mone he apy o psycho he apy we e excluded. To add ess he second aim, s udies ha measu ed dep essi e symp oma ology h ough a psychome ically alida ed ins umen and ha p esen ed a condi ion o bio- o neu o eedback in a an- domized con olled ial we e conside ed o inclusion. The objec i e o including his second laye o s udies was o de e - mine he ex en o which dep essi e symp oms a e ea ed h ough bio- and neu o eedback echniques in o he men al dis- o de s and pa icula ly in medical s udies. In o he wo ds, all s udies assessing dep essi e symp oma ology as a seconda y ou - come measu e we e comp ised. Unpublished s udies, con e ence pape s and p oceedings, he- sis and a icles published in non-pee - e iewed we e excluded om he s udy selec ion in bo h sea ches. S udy selec ion p ocedu e One e iewe comple ed all da abase sea ches o bo h objec i es a he same ime. All esul s we e expo ed o EndNo e and dupli- ca es we e elimina ed. A e ha , wo e iewe s (JFA and DC) sc eened independen ly all i les and abs ac s o iden i y po en- ially ele an a icle o any o he wo objec i es. Two di e en olde s we e c ea ed, one o each objec i e. F om he o al amoun o s udies ha we e included o u he examina ion, he wo independen e iewe s ead ull ex s o de e mine i he eligibili y c i e ia we e ul illed. Disag eemen s we e esol ed 2 J. Fe nández‐Al a ez e al. h ps://doi.o g/10.1017/S0033291721004396 Downloaded om h ps://www.camb idge.o g/co e. Uni e si a Jaume I, on 12 Jan 2022 a 08:03:41, subjec o he Camb idge Co e e ms o use, a ailable a h ps://www.camb idge.o g/co e/ e ms. h ough discussion, and i necessa y, a hi d e iewe was consul ed. Quali y assessmen o s udies Coch ane Collabo a ion Risk o Bias ool has been used o assess sou ces o bias in andomized con olled ials (RCTs). Ou con- side ed c i e ia en ail lack o alloca ion/concealmen , lack o blinding, incomple e accoun ing o ou come o pa ien e en s, and selec i e ou come epo ing. E ec size calcula ion and coding o s udies We es ima ed he e ec size o bo h he di e ence in change be ween he g oups as well as he p e-pos change wi hin he bio- eedback g oups by using Hedges’g, a a ia ion o Cohen’sd which akes in o accoun o biases associa ed wi h small sample sizes (Hedges & Olkin, 1985). When he g oup mean, s anda d de ia ion (S.D.), a iance o s anda d e o o he mean, and a numbe o subjec s we e a ailable o each g oup, hese da a we e p e e ably used o calcula e he e ec size. When some o hese da a we e missing, we looked o o he da a allowing o he e ec size compu a ion, such as uns anda dized mean di e - ences, and p alues. I mul iple measu emen s o dep ession we e used in he same s udy, a pooled e ec size was calcula ed in o de o include in he me a-analyses only a single e ec size o each s udy. The pooled es ima e o he e ec size was calcu- la ed as he a e age o he di e en e ec sizes o each measu e. The a iance o he pooled es ima e was also calcula ed as he a e age o he di e en a iances o he e ec sizes; as he co ela- ions among he measu es a e o en no epo ed in he pape s, such app oach ep esen s he mos conse a i e way o calcula e he pooled e ec size a iance (i.e. he s a egy ha leads o he la ges a iances o he es ima e o he pool e ec size, assuming a pe ec co ela ion among measu es). Simila ly, a pooled e ec size was calcula ed and included in he mea -analysis i ei he mul iple bio eedback o con ol g oups we e used in he s udy. In addi ion, he ollowing s udy cha ac e is ics we e coded and included in he analyses as mode a o s: (1) sex (% o emale sub- jec s in he con ol g oup); (2) age (mean age in he con ol g oup), (3) leng h o ea men (numbe o sessions), (4) ype o bio eedback in e en ion (hea a e a iabili y bio eedback o neu o eedback) (5) yea o publica ion, (6) expe imen al design ( andomized-con ol ial), (7) me hodological quali y o s udies. Coding o mode a o s 1–3 may be no possible o all s udies. In such case, a “no a ailable”was be inse ed. Two o he au ho s independen ly pe o med he compu a ion o e ec sizes and any disc epancy was esol ed be o e analysis, wi h he in ol emen o a hi d au ho in case o pe sis en disag eemen . Me a-analy ic s a is ics Each s udy e ec size was weigh ed by i s in e se a iance ( he sum o he wi hin-s udy a iance and an es ima e o he be ween- s udies a iance), gi ing a la ge weigh ing o s udies wi h la ge sample sizes han hose wi h small sample sizes. Be o e excluding a s udy because i was no possible o calcula e he e ec size due o a lack o enough s a is ical de ails epo ed in he pape , we ied o con ac he co esponding au ho (only o s udies published less han 10 yea s ago) asking o he missing de ails. O he wise, he s udy was excluded om he analyses. The pooled e ec sizes we e es ima ed using andom-e ec s models (Res ic ed Maximum-Likelihood Es ima ion), wi h con i- dence in e als and s a is ical es calcula ed wi h Knapp–Ha ung me hod (Knapp & Ha ung, 2003), which hypo hesizes po en ial signi ican he e ogenei y among s udies. A signi icance le el o 0.05 was applied. Q s a is ic and I 2 index we e used o in es iga e he he e ogen- ei y o he e ec sizes among s udies. The signi icance le el o he Q s a is ic was se a 0.1 o adjus o he limi ed s a is ical powe o his es (Pe i i, 2001). I 2 can be in e p e ed as he pe cen age o he o al a iabili y in a se o e ec sizes ha canno be a ib- u ed only o he sampling e o wi hin s udies. I he Qs a is ic esul ed signi ican o I 2 sugges ed he e ogenei y, we checked whe he he sou ce o his he e ogenei y migh ha e been a ibu- ed o one single e ec size ou lying om all o he s. In his case, we epea ed he me a-analy ic analyses one-by-one emo ing each e ec size. The e ec size was conside ed ou lying when i s exclu- sion om he analyses yielded a esolu ion o he he e ogenei y, and i s inclusion du ing he emo al o o he e ec sizes did no . Finally, mode a o s we e included in he me a-analysis in o de o y explaining possible sou ces o he e ogenei y among he e ec sizes. Mode a o s we e conside ed only i hey we e a ailable in a leas ou independen s udies. Conside ing he lim- i ed numbe o s udies expec ed o be included in he cu en me a-analyses, mode a o analysis will be pe o med sepa a ely o each mode a o . Publica ion bias Publica ion bias was assessed by isual inspec ion o he unnel plo s and by he Egge ’s eg ession es (one- ailed po <0.05 was conside ed o indica e he p esence o he bias) (Egge , Da ey Smi h, Schneide , & Minde , 1997). We also used he im-and- ill me hod om Du al and Tweedie (Pe e s, Su on, Jones, Ab ams, & Rush on, 2007) o de e mine he na u e o po en ial publica ion bias and o compu e an es ima ed e ec size ha accoun s o i . Resul s Included s udies As illus a ed in he PRISMA low cha (Fig. 1), a o al o 11 786 eco ds ha e been e ie ed om he ini ial da abase sea ches. A e emo ing all duplica ed a icles, he i s sc eening s ep (examina ion o i les and abs ac s) iden i ied 7235 e e ences ha we e o po en ial in e es o ou me a-analysis. Such a p ocess was ca ied ou by wo e iewe s and yielded 18 and 24 e e ences, o each o he espec i e s eps o ou s udy. A o al numbe o 22 pape s ul illed he inclusion c i e ia and we e inally included in he s udy. 24 pape s we e excluded because hey did no sa is y he inclusion c i e ia (desc ibed in Fig. 1). The whole p ocedu e was independen ly done by wo e iewe s (JFA and DC). A lowcha o he gene al inclusion p ocedu e is epo ed in Fig. 1. No eply was ecei ed om any au ho we con ac ed o ob ain missing da a. Desc ip ions o all he included s udies wi h ele an a iables and s udy-le el cha ac e is ics coded o each s udy a e epo ed in Tables 1–3, o each s ep o he s udy. Psychological Medicine 3 h ps://doi.o g/10.1017/S0033291721004396 Downloaded om h ps://www.camb idge.o g/co e. Uni e si a Jaume I, on 12 Jan 2022 a 08:03:41, subjec o he Camb idge Co e e ms o use, a ailable a h ps://www.camb idge.o g/co e/ e ms. E icacy o bio- and neu o eedback o MDD (le el 1) P e-pos be ween-g oup e ec sizes Fo he p e-pos be ween-g oup analysis compa ing he bio- and neu o eedback and con ol g oups, he andom-e ec s analyses yielded an o e all e ec size o Hedges’g= 0.717 (95% CI 0.2121–1.1224, = 3.357, p= 0.0121) (Fig. 2), indica ing a g ea e e icacy o bio- and neu o eedback compa ed o con ol ea - men s in he ea men o MDD. In o al, his me a-analysis was based on eigh s udies and 176 pa ien s. No e idence o sig- ni ican he e ogenei y was ound conside ing he Q s a is ics (Q= 8.193, p= 0.316), while I 2 esul ed o 29%, indica ing a small po en ial he e ogenei y among he e ec sizes o he single s udies (Higgins, 2003). Mode a o analyses and publica ion bias The e ec o all mode a o s esul ed s a is ically non-signi ican (see Table 4). The occu ence o publica ion bias was no sugges ed by any o he es s used (T im and Fill analysis) sugges s ha no s udies needed o all o he igh o le o he mean o make he plo symme ical, and Egge ’s es esul ed no signi i- can ( p= 0.4365) as well as by isual inspec ion o he unnel plo (Fig. 3). P e-pos wi hin-g oup e ec sizes Fo he p e-pos wi hin-g oup analysis o he sole bio eedback ea men , he andom e ec s me a-analysis yielded an o e all wi hin-g oup e ec size o Hedges’g= 1.050 (95% CI 0.492– 1.608, = 5.991, p= 0.001) (Fig. 4), which indica es a signi ican e icacy o bio- and neu o eedback in imp o ing MDD symp om- a ology (n= 110). No e idence o signi ican he e ogenei y was ound conside ing he Qs a is ics (Q= 3.353, p= 0.340), and also I 2 (13%) sugges s a e y limi ed he e ogenei y among he e ec sizes o he single s udies. Fig. 1. Flowcha o included s udies in he me a-analysis. 4 J. Fe nández‐Al a ez e al. h ps://doi.o g/10.1017/S0033291721004396 Downloaded om h ps://www.camb idge.o g/co e. Uni e si a Jaume I, on 12 Jan 2022 a 08:03:41, subjec o he Camb idge Co e e ms o use, a ailable a h ps://www.camb idge.o g/co e/ e ms. Table 1. Be ween e ec sizes o Neu o- and bio eedback o dep essi e symp oma ology in all condi ions Coun y Type o BF Con ol g oup Design Mean age BF Mean age CG Sample size (BF/CG) %o emale BF %o emale CG Ins umen N . o sessions Risk o bias (A/C; LB; IA; SOR) Young e al. (2017) USA MRI Placebo RCT/Double-Blind, placebo-con olled 32 31 36 (19/17) 66% 75% Composi e o 3 scales (HDRS; MADRS; BDI-II) 2 +/+/+/+ Choi e al. (2011) Sou h Ko ea EEG Placebo RCT/Single-blind 28.46 28.54 23 (12/11) 20% 57% Composi e o 2 scales (HDRS & BDI-II) 5 + /U/ + / + Li e al. (2015) China/USA MRI Regula ehabili a ion a RCT 54.38 59.64 24 (13/11) 62% 27% HDRS 3 imes/w. o 4–6w U/U/U/ + Caldwell and S e en (2018) USA HRV Psycho he apy ac i e RCT 20.09 20.64 20 (10/10) 100% 100% Beck Dep ession In en o y-II 5 U/U/U/ + Yuan e al. (2014) USA MRI NF based on Le HIPS Case–con ol 38 35 27 (13/14) 79% 85% HDRS 1 + /U/ + / + Linden e al. (2012) UK/ Ne he lands MIR & EEG Heal hy subjec s / image y p ocedu e Expe imen al design 48.37 48.5 16 (8/8) 0% 37% HDRS-17 4 U/U/U/ + Zo e e al. (2016) USA MRI Placebo RANDOMIZED/ Double-Blind, placebo-con olled 41 34 13 69% 85% P o ile o Mood S a es 2 + /U/ + / + Mehle e al. (2018) Ne he lands MRI Ac i a ion o highe isual p ocesses .RCT 47.19 46.94 32 (16/16) 69% 62,5% HDRS-17 5 + / + / + / + A/C, Alloca ion/concealmen ; Beck Dep ession In en o y–II; BF, bio eedback; CG, Con ol g oup; EEG, elec oencephalog am; MRI, unc ional magne ic esonance imaging; HAD, Hamil on Dep ession Ra ing Scale; HIPS, ho izon al segmen o in apa ie al sulcus; HRV, Hea a e a iabili y; IA, Incomple e accoun ing o ou come o pa ien e en s; LB, Lack o blinding; MADRS, Mon gome y-Åsbe g Dep ession; RCT, andomized con olled ial; N . o sessions, Numbe o sessions; SOR, Selec i e Ou come Repo ing. aHamil on Dep ession Ra ing Scale (17-i ems). Psychological Medicine 5 h ps://doi.o g/10.1017/S0033291721004396 Downloaded om h ps://www.camb idge.o g/co e. Uni e si a Jaume I, on 12 Jan 2022 a 08:03:41, subjec o he Camb idge Co e e ms o use, a ailable a h ps://www.camb idge.o g/co e/ e ms. Mode a o analyses Among he mode a o s (see Table 5), only he expe imen al design esul ed in ha ing a signi ican e ec in mode a ing he o e all wi hin-g oup e ec size (F= 126.582, p= 0.008). The wi hin-g oup e ec size o bio eedback ea men s esul ed signi ican bo h in andomized-con olled s udies (Hedges’g = 1.391, 95% CI 1.216– 1.566, = 34.164, p< 0.001) and in non- andomized s udies (Hedges’g= 0.783, 95% CI 0.630–1.566, =22.045, p=0.002), wi h he la e esul ing signi ican ly g ea e han he o me . Publica ion bias T im and Fill analysis indica ed ha one s udy would need o all o he le o he mean o make he plo symme ical, while no s udies on he o he side. This sugges s ha he o e all e ec size calcula ed in he wi hin-g oup analysis may be in la ed by he lack o inclusion in he me a-analysis o some un epo ed s udy, as i is also e idenced by isual inspec ion o he unnel plo (Fig. 5). Howe e , he andom-e ec s me a-analysis pe o med adjus ing o missing s udies s ill yielded a signi ican o e all e ec size (Hedges’g= 1.196, 95% CI 0.985–1.407, = 11.102, p< 0.0001), wi h only a e y small educ ion o i s p e ious magni ude. Ins ead, no e idence o publica ion bias was sugges ed by he Egge ’s es ( p= 0.281) E icacy o bio eedback o dep essi e symp oms in o he condi ions (le el 2) P e-pos be ween-g oup e ec sizes Fo he p e-pos be ween-g oups analysis compa ing he bio- and neu o eedback and con ol g oups, he andom-e ec s analyses yielded a signi ican o e all e ec size o Hedges’g= 0.303 (95% CI 0.121–0.484, = 2.217, p= 0.003) (Fig. 6), indica ing a g ea e e icacy o bio- and neu o eedback compa ed o con ol ea - men s o dep essi e symp oms (n= 736). No e idence o signi i- can he e ogenei y was ound conside ing he Qs a is ics (Q= 4.350, p= 0.993), and also I 2 (0%) sugges s a lack o he e ogenei y among he e ec sizes o he single s udies. Mode a o analyses and publica ion bias The e ec o all mode a o s esul ed s a is ically non-signi ican (Table 6) and no occu ence o publica ion bias was sugges ed by any o he es s used (T im and Fill analysis sugges s ha no s udies needed o all o he igh o le o he mean o make he plo symme ical, and Egge ’s es esul ed no signi ican wi h a p= 0.911) as well as by isual inspec ion o he unnel plo (Fig. 7). Table 2. Wi hin e ec sizes o Neu o- and bio eedback o dep essi e symp oma ology in all condi ions Coun y Type o BF Design Mean age Sample size Pe cen emale Ins umen N . o sessions Risk o bias (A/ C; LB; IA; SOR) Young e al. (2017) USA MRI RANDOMIZED/ Double-Blind, placebo-con olled 32 19 66%/ 75% Composi e o 3 scales (HDRS; MADRS; BDI-II) 2+/+/ +/+ Choi e al. (2011) Sou h Ko ea EEG RANDOMIZED/ Single-blind 28.46 12 20%/ 57% Composi e o 2 scales (HDRS & BDI-II) 5 + /U/ +/+ Yuan e al. (2014) USA MRI Case–con ol 38 13 79%/ 85% Hamil on Dep ession Ra ing Scale 1 + /U/ +/+ Zo e e al. (2016) USA MRI RANDOMIZED/ Double-Blind, placebo-con olled 41 13 69%/ 85% P o ile o Mood S a es 2 + /U/ +/+ A/C, Alloca ion/concealmen ; BF, bio eedback; CG, Con ol g oup; EEG, elec oencephalog am; MRI, unc ional magne ic esonance imaging; IA, Incomple e accoun ing o ou come o pa ien e en s; LB, Lack o blinding; RCT, andomized con olled ial; N . o sessions, Numbe o sessions; SOR, Selec i e Ou come Repo ing. Fig. 2. P e-pos be ween-g oup e ec sizes in le el 1. 6 J. Fe nández‐Al a ez e al. h ps://doi.o g/10.1017/S0033291721004396 Downloaded om h ps://www.camb idge.o g/co e. Uni e si a Jaume I, on 12 Jan 2022 a 08:03:41, subjec o he Camb idge Co e e ms o use, a ailable a h ps://www.camb idge.o g/co e/ e ms. Table 3. Be ween e ec sizes o Neu o- and bio eedback o dep essi e symp oma ology in all condi ions Coun y Type o BF Compa ison condi ion Popula ion Sample size Pe cen emale Ins umen N . o sessions Risk o bias (A/C; LB; IA; SOR) Hallman e al. (2011) Sweden HRV Con ol (check i WL) Ch onic neck pain 23 (HRV 12; CG 11) 91% HADS 10 U/U/ + / + Pa on e al. (2013) I aly HRV Daily counseling sessions A e ca diac su ge y pa ien s 26 (HRV 13; CG 13) 22.5% CES-D 5 (45’) U/U/U/ + Penzlin e al. (2015) Ge many/ USA HRV CBT Addic ion (alcohol) 43 (HRV 24; CG 19) ¿?¿? BDI-II 6 (20’)+/+/+/+ Ra anasi ipong e al. (2015) USA/ Thailand HRV WL Heal hy popula ion 60 (HRV 30; CG 30) 97% CES-D 1 a + /U/U/ + Swanson e al. (2009) USA HRV Quasi- alse alpha- he a bio eedback Hea Failu e 29 (HRV 15; CG 14) 80% CES-D 6 (45’) + /U/ + / + Zucke e al. (2009) USA HRV P og essi e muscle elaxa ion PTSD 38 (HRV 19; CG 19) 44.7% BDI-II 4 weeks once a day o 20’a home + / + /U/ + Schönbe g e al. (2017) Ge many Neu o Sham neu o eedback ADHD 75 (NBF 37; CG 38) 44% BDI-II + / + / + / + Schönbe g e al. (2017) Ge many Neu o Me a-cogni i e g oup he apy g oup ADHD 75 (NBF 37; CG 38) 44% BDI-II 30 + / + / + / + Hsueh e al. (2016) Taiwan Neu o Sham neu o eedback Memo y enhancem. 25; 25 60% BDI-II 12 + / + /U/ + Lackne e al. (2016) Aus ia / UK Neu o T aining sessions (di e en psycho he apeu ic aspec s) Addic ion (alcohol) 25 (NF 13; CG 12) 0% BDI-V (modi ied e sion o BDI) 12 + / + /U/ + Menella e al. (2017) I aly Neu o Ac i e con ol aining Heal hy popula ion 32 (NF 16; CG 16) 100% BDI-II 7 (45’) + / + /U/ + Dehghani-A ani e al. (2013) I an Neu o TAU (medica ion) Addic ion (opia e) 20 (NF 10; CG 10) No in o GHQ-28 30 + / + /U/ + Choob o oushzadeh e al. (2015) I an Neu o TAU (medica ion) Mul iple schle osis 24 (NF 12; CG 12) 50% HADS 16 + / + /U/ + Yu e al. (2018) Taiwan HRV Medicine Co ona y A e y Disease 210 (105 HRV; 105 CG) 12% HRV 10.20%CG BDI-II 6 (60’) + /U/ + / + A/C, Alloca ion/concealmen ; ADHD, A en ion De ici Hype ac i i y Diso de ; BDI, Beck Dep ession In en o y; CES-D, Cen e o Epidemiologic S udies Dep ession Scale; CG, Con ol g oup; GHQ, Gene al Heal h Ques ionnai e; HADS, Hospi al Anxie y and Dep ession Scale; HRV, hea a e a iabili y; IA, Incomple e accoun ing o ou come o pa ien e en s; PTSD, Pos T auma ic S ess Diso de ; RCT, andomized con olled ial; NF, Neu o eedback; N . o sessions, Numbe o sessions; LB, Lack o blinding; SOR, Selec i e Ou come Repo ing TAU, T ea men as usual; Wai ing Lis . a One session wi h he apis s and hen ou weeks o use a home h ee imes a day. Psychological Medicine 7 h ps://doi.o g/10.1017/S0033291721004396 Downloaded om h ps://www.camb idge.o g/co e. Uni e si a Jaume I, on 12 Jan 2022 a 08:03:41, subjec o he Camb idge Co e e ms o use, a ailable a h ps://www.camb idge.o g/co e/ e ms. P e-pos wi hin-g oup e ec sizes P e-pos wi hin-g oup analysis o he sole bio eedback ea men s could no be pe o med because only one s udy (Kayi an, Du sun, Du sun, E mu lu, & Ka amü sel, 2010) sa is ied all he inclusion c i e ia o his pa o he me a-analysis. Discussion To ou knowledge, his is he i s me a-analysis o bio- and neu- o eedback echniques o he ea men o MDD and dep essi e symp oms. Taken oge he , hese indings sugges ha bio- and neu o eedback cons i u e e ec i e in e en ions o bo h indi i- duals wi h clinical dep ession and wi h seconda y dep essi e symp oma ology. Besides, he esul s a e in line wi h he indings o p e ious quali a i e e iews (Hammond, 2005; Linden, 2014; Sacche & Go lib, 2016; Young e al., 2018). Bio- and neu o eedback o MDD While he wi hin-g oup analyses yielded an e ec size o Hedges’ g=0.717, he be ween-g oup analysis e ealed an e ec size o Hedges’g=1.050. The mode a o analyses indica e ha ea men e icacy is only signi ican when accoun ing o expe imen al design, in a o o RCTs in compa ison o non-RCTs. The ac ha all o he mode a o s we e non-signi ican may indica e ha esul s can be gene alized o a ange o social and clinical cha ac e is ics. Ne e heless, hese esul s mus be aken wi h cau ion gi en ha only one RCT has been conduc ed o MDD using HRVB (Caldwell & S e en, 2018), and hus making i di icul o con- clude ha his echnique is e ec i e o MDD. Addi ionally, many o he iden i ied s udies a ge ing clinical dep ession wi h HRVB could no be included due o se e al easons, such as he absence o a con ol g oup o he ac ha pa icipan s we e aking an idep essan s. So a , a conside able numbe o s udies ha e p e- iously indica ed ha HRVB is e ec i e o dep ession, equen ly e e ing o he esea ch conduc ed by Ka a idas e al. (2007)o Siepmann, Aykac, Un e dö e , Pe owski, and Mueck-Weymann (2008). While i is ue ha bo h in e en ions we e e ec i e, none o hem had ei he a con ol condi ion, in bo h cases pa ien s we e unde psychopha macological ea men and he samples we e unde powe ed. Con e sely, neu o eedback ga he ed mo e e idence. The exis - ing s udies o neu o eedback comp ised bo h s udies b ough o h o EEG (Cheon, Koo, & Choi, 2016; Choi e al., 2011) and MRI (Li e al., 2015; Linden e al., 2012; Mehle e al., 2018; Young e al., 2014,2017). Among he EEG s udies, he a - ge was o egula e he on al asymme ic ac i a ion inc easing he ela i e le on al ac i i y and consequen ly o imp o e dep essi e symp oma ology. In he case o he MRI s udies, some o hem a ge ed he amygdala (Young e al., 2014,2017) and o he s he insula and la e al p e on al a eas (Mahle e al., 2018), all o which a e in ol ed in he egula ion o emo ions (Sebas ian & Ahmed 2018). F om his poin o iew, emo ion egula ion is a well-es ablished ansdiagnos ic ac o ha explains bo h he appea ance and main enance o a as a ay o a ec i e diso de s (Aldao, Nolen-Hoeksema, & Schweize , 2010), includ- ing MDD and dep essi e symp oms. In eg a ing he e idence Fig. 3. Funnel plo be ween analyses in le el 1. Fig. 4. P e-pos wi hin-g oup e ec sizes in le el 1. 8 J. Fe nández‐Al a ez e al. h ps://doi.o g/10.1017/S0033291721004396 Downloaded om h ps://www.camb idge.o g/co e. Uni e si a Jaume I, on 12 Jan 2022 a 08:03:41, subjec o he Camb idge Co e e ms o use, a ailable a h ps://www.camb idge.o g/co e/ e ms. Table 4. Mode a o s be ween analysis le el 1 INTERCEPT MODERATOR MODERATOR In e p e a ion o he es ima e Es ima e SE p 95% CI lowe bound 95% CI highe bound In e p e a ion o he es ima e Es ima e SE al p al 95% CI lowe bound 95% CI highe bound Type o eedback Es ima ed e ec size o h bio eedback 0.996 0.591 1.686 0.143 −0.450 2.443 Es ima ed e ec size di e ence be ween neu o eedback and h bio eedback −0.321 0.640 −0.503 0.633 −1.886 1.243 Randomized con ol ial Es ima ed e ec size o non- andomized con ol ials 0.958 0.322 2.973 0.025 0.170 1.747 Es ima ed e ec size di e ence be ween andomized and non- andomized con ol ials −0.452 0.430 −1.051 0.334 −1.503 0.600 Yea o publica ion Es ima ed e ec size o s udies published in 2011 1.459 0.406 3.593 0.011 0.465 2.452 Es ima ed change o he e ec size o e e y yea o publica ion a e 2011 −0.167 0.076 −2.214 0.069 −0.352 0.018 Numbe o sessions Es ima ed e ec size wi h 4 sessions 0.717 0.228 3.144 0.020 0.159 1.276 Es ima ed change e ec size change o e e y addi ional session 0.025 0.066 0.379 0.718 −0.136 0.186 A e age age o bio eedback g oup Es ima ed e ec size a a e age age o 40 yea s 0.724 0.230 3.144 0.020 0.161 1.288 Es ima ed e ec size change o e e y addi ional yea o he mean age 0.003 0.025 0.109 0.917 −0.059 0.064 Pecen age o emale subjec s in he bio eedback g oup Es ima ed e ec size wi h only male subjec s 1.473 0.419 3.518 0.013 0.448 2.498 Es ima ed e ec size di e ence be ween only emale and only male subjec s −1.362 0.651 −2.092 0.081 −2.954 0.231 Quali y o he s udy Es ima ed e ec size wi h low isk o bias 0.210 0.284 0.741 0.487 −0.485 0.906 Es ima ed e ec size di e ence e ec size be ween unknown/high and low isk o bias 0.738 0.369 1.997 0.093 −0.166 1.642 Psychological Medicine 9 h ps://doi.o g/10.1017/S0033291721004396 Downloaded om h ps://www.camb idge.o g/co e. Uni e si a Jaume I, on 12 Jan 2022 a 08:03:41, subjec o he Camb idge Co e e ms o use, a ailable a h ps://www.camb idge.o g/co e/ e ms. Pe na,G.,G assi,M.,Caldi ola,D.,&Neme o ,C.B.(2018).The e olu iono pe sonalized psychia y: Will echnology make i happen soone ? Psychological Medicine,48(5), 705–713. h ps://doi.o g/10.1017/S0033291717002859. Pe e s, J. L., Su on, A. J., Jones, D. R., Ab ams, K. R., & Rush on, L. (2007). Pe o mance o he im and ill me hod in he p esence o publica ion bias and be ween-s udy he e ogenei y. S a is ics in Medicine,26(25), 4544–4562. h ps://doi.o g/10.1002/sim.2889. Pe i i, D. B. (2001). App oaches o he e ogenei y in me a-analysis. S a is ics in Medicine,20(23), 3625–3633. h ps://doi.o g/10.1002/sim.1091. Po ges, S. W. (2007). The poly agal pe spec i e. Biological Psychology,74(2), 116–143. h ps://doi.o g/10.1016/J.BIOPSYCHO.2006.06.009. P ice, J. L., & D e e s, W. C. (2010). Neu oci cui y o mood diso de s. Neu opsychopha macology,35(1), 192–216. h ps://doi.o g/10.1038/npp.2009.104. Rami ez, R., Palencia-Le le , M., Gi aldo, S., & Vam akousis, Z. (2015). Musical neu o eedback o ea ing dep ession in elde ly people. F on ie s in Neu oscience,9(OCT), 1–10. h ps://doi.o g/10.3389/ nins.2015.00354. Raskind, M. A. (2008). Diagnosis and ea men o dep ession como bid wi h neu ologic diso de s. The Ame ican Jou nal o Medicine,121(11), S28–S37. h ps://doi.o g/10.1016/j.amjmed.2008.09.011. Ra anasi ipong, P., Kaewboonchoo, O., Ra anasi ipong, N., Hanklang, S., & Chumchai, P. (2015). Bio eedback in e en ion o s ess, anxie y, and dep ession among g adua e s uden s in public heal h nu sing. Nu sing Resea ch and P ac ice,1–5. doi: 10.1155/2015/160746. Repe o, C., Gaggioli, A., Palla icini, F., Cip esso, P., Raspelli, S., & Ri a, G. (2013). Vi ual eali y and mobile phones in he ea men o gene alized anxie y diso de s: A phase-2 clinical ial. Pe sonal and Ubiqui ous Compu ing,17(2), 253–260. h ps://doi.o g/10.1007/s00779-011-0467-0. Rosen eld, J. P. (2000). An EEG bio eedback p o ocol o a ec i e diso de s. Clinical EEG and Neu oscience,31(1), 7–12. h p://dx.doi.o g/10.1177/ 155005940003100106. Sacche , M. D., & Go lib, I. H. (2016). Recen de elopmen s and u u e di ec- ions. Expe Re iew o Neu o he apeu ics,16(9), 1003–1005. h ps://doi.o g/ 10.1111/2041-210X.12. Schoelle , F., Be and, P., Ge y, L. J., Jain, A., Ho owi z, A. H., & Zenasni, F. (2019). Combining i ual eali y and bio eedback o os e empa hic abil- i ies in humans. F on ie s in Psychology,9(Feb ua y), 1–5. h ps://doi.o g/ 10.3389/ psyg.2018.02741. Schoenbe g, P. L. A., & Da id, A. S. (2014). Bio eedback o psychia ic diso - de s: A sys ema ic e iew. Applied Psychophysiology and Bio eedback,39(2), 109–135. h ps://doi.o g/10.1007/s10484-014-9246-9. Schönenbe g, M., Wiedemann, E., Schneid , A., Schee , J., Logemann, A., Keune, P. M., & Hau zinge , M. (2017). Neu o eedback, sham neu o eed- back, and cogni i e-beha iou al g oup he apy in adul s wi h a en ion-de - ici hype ac i i y diso de : A iple-blind, andomised, con olled ial. The Lance Psychia y,4(9), 673–684. doi: 10.1016/S2215-0366(17)30291-2. Sebas ian, C. L., & Ahmed, S. P.. (2018). Neu obiology o emo ion egula ion. In An hony R. B., Adam J. C., Ru h E. M., & Pia R. (Eds.), The Wiley Blackwell Handbook o Fo ensic Neu oscience (Vol. 1, pp. 125–140). Wiley Blackwell. Siepmann, M., Aykac, V., Un e dö e , J., Pe owski, K., & Mueck-Weymann, M. (2008). A pilo s udy on he e ec s o hea a e a iabili y bio eedback in pa ien s wi h dep ession and in heal hy subjec s. Applied Psychophysiology and Bio eedback,33(4), 195–201. h ps://doi.o g/10.1007/ s10484-008-9064-z. Swanson, K. S., Ge i z, R. N., B own, M., Spi a, J., Gua ne i, E., & S ole niy, L. (2009). The e ec o bio eedback on unc ion in pa ien s wi h hea ail- u e. Applied Psychophysiology and Bio eedback,34(2), 71–91. doi: 10.1007/ s10484-009-9077-2. Thaye , J. F., & Lane, R. D. (2009). Claude Be na d and he hea -b ain con- nec ion: Fu he elabo a ion o a model o neu o isce al in eg a ion. Neu oscience & Biobeha io al Re iews,33(2), 81–88. h ps://doi.o g/10. 1016/j.neubio e .2008.08.004. Thaye , J. F., & Ma he , M. (2018). How hea a e a iabili y a ec s emo ion egula ion b ain ne wo ks. Cu Opin Beha Sci.,19,98–104. doi: 10.1016/ J.COBEHA.2017.12.017. Thibaul , R. T., Li shi z, M., & Raz, A. (2016). The sel - egula ing b ain and neu o eedback: Expe imen al science and clinical p omise. Co ex,74, 247–261. doi: 10.1016/j.co ex.2015.10.024. Thibaul , R. T., & Raz, A. (2017). The psychology o neu o eedback: Clinical in e en ion e en i applied placebo. Ame ican Psychologis ,72(7), 679– 688. doi: 10.1037/amp0000118. Thibaul , R. T., & Raz, A. (2018). A consensus amewo k o neu o eedback esea ch (and he pe ils o un ounded neu o educ ionism): Reply o Micoulaud-F anchi and Fo e (2018). Ame ican Psychologis ,73(7), 936– 937. doi: 10.1037/amp0000366. Wa son, D. (2009). Di e en ia ing he mood and anxie y diso de s: A quad i- pa i e model. Annual Re iew o Clinical Psychology,5(1), 221–247. h ps:// doi.o g/10.1146/annu e .clinpsy.032408.153510. WHO. (2017). Dep ession and o he common men al diso de s: Global heal h es ima es. Wo ld Heal h O ganiza ion. Young, K. D., Siegle, G. J., Zo e , V., Phillips, R., Misaki, M., Yuan, H., … Bodu ka, J. (2017). Randomized clinical ial o eal- ime MRI amygdala neu o eedback o majo dep essi e diso de : E ec s on symp oms and au obiog aphical memo y ecall. Ame ican Jou nal o Psychia y,174(8), 748–755. h ps://doi.o g/10.1176/appi.ajp.2017.16060637. Young, K. D., Zo e , V., Phillips, R., Misaki, M., D e e s, W. C., & Bodu ka, J. (2018). Amygdala eal- ime unc ional magne ic esonance imaging neu o- eedback o majo dep essi e diso de : A e iew. Psychia y and Clinical Neu osciences,72(7), 466–481. h ps://doi.o g/10.1111/pcn.12665. Young, K. D., Zo e , V., Phillips, R., Misaki, M., Yuan, H., D e e s, W. C., & Bodu ka, J. (2014). Real- ime MRI neu o eedback aining o amygdala ac i i y in pa ien s wi h majo dep essi e diso de . PLoS ONE,9(2). h ps://doi.o g/10.1371/jou nal.pone.0088785. Yu, L.-C., Lin, I.-M., Fan, S.-Y., Chien, C.-L., & Lin, T.-H. (2018). One-yea ca dio ascula p ognosis o he andomized, con olled, sho - e m hea a e a iabili y bio eedback among pa ien s wi h co ona y a e y disease. In e na ional Jou nal o Beha io al Medicine,25(3), 271–282. doi: 10.1007/s12529-017-9707-7. Yuan, H., Young, K. D., Phillips, R., Zo e , V., Misaki, M., & Bodu ka, J.. (2014). Res ing-s a e unc ional connec i i y modula ion and sus ained changes a e eal- ime unc ional magne ic esonance imaging neu o- eedback aining in dep ession. B ain Connec i i y,4(9), 690–701. doi: 10.1089/b ain.2014.0262. Zo e , V., Yuan, H., Misaki, M., Phillips, R., Young, K. D., Feldne , M. T., & Bodu ka, J. (2016). Co ela ion be ween amygdala BOLD ac i i y and on al EEG asymme y du ing eal- ime MRI neu o eedback aining in pa ien s wi h dep ession. Neu oImage: Clinical,11, 224–238. doi: 10.1016/ j.nicl.2016.02.003. Zucke , T. L., Samuelson, K. W., Muench, F., G eenbe g, M. A., & Ge i z, R. N. (2009). The e ec s o espi a o y sinus a hy hmia bio eedback on hea a e a iabili y and pos auma ic s ess diso de symp oms: A pilo s udy. Applied Psychophysiology and Bio eedback,34(2), 135–143. Zuelke, A. E., Luck, T., Sch oe e , M. L., Wi e, A. V., Hinz, A., Engel, C., … Riedel-helle , S. G. (2018). The associa ion be ween unemploymen and dep ession – esul s om he. Jou nal o A ec i e Diso de s,235(Ma ch), 399–406. h ps://doi.o g/10.1016/j.jad.2018.04.073. 16 J. Fe nández‐Al a ez e al. h ps://doi.o g/10.1017/S0033291721004396 Downloaded om h ps://www.camb idge.o g/co e. Uni e si a Jaume I, on 12 Jan 2022 a 08:03:41, subjec o he Camb idge Co e e ms o use, a ailable a h ps://www.camb idge.o g/co e/ e ms.