RESEARCH ARTICLE Open Access
Clinical ials in pallia i e ca e: a sys ema ic
e iew o hei me hodological
cha ac e is ics and o he quali y o hei
epo ing
Raquel Bouça-Machado
1
, Madalena Rosá io
1
, Joana Ala cão
2
, Leono Co eia-Guedes
1
, Daisy Ab eu
1
and Joaquim J. Fe ei a
1,3*
Abs ac
Backg ound: O e he pas decades he e has been a signi ican inc ease in he numbe o published clinical ials
in pallia i e ca e. Howe e , empi ical e idence sugges s ha he e a e me hodological p oblems in he design and
conduc o s udies, which aises ques ions abou he alidi y and gene alisabili y o he esul s and o he s eng h o
he a ailable e idence. We sough o e alua e he me hodological cha ac e is ics and assess he quali y o epo ing o
clinical ials in pallia i e ca e.
Me hods: We pe o med a sys ema ic e iew o published clinical ials assessing he apeu ic in e en ions in
pallia i e ca e. T ials we e iden i ied using MEDLINE ( om i s incep ion o Feb ua y 2015). We assessed me hodological
cha ac e is ics and desc ibe he quali y o epo ing using he Coch ane Risk o Bias ool.
Resul s: We e ie ed 107 s udies. The mos common medical ield s udied was oncology, and 43.9% o ials e alua ed
pha macological in e en ions. Symp om con ol and physical dimensions (e.g. in e en ion on pain, b ea hlessness,
nausea) we e he pallia i e ca e-speci ic issues mos s udied. We ound unde - epo ing o key in o ma ion in pa icula on
andom sequence gene a ion, alloca ion concealmen , and blinding.
Conclusions: While he numbe o clinical ials in pallia i e ca e has inc eased o e ime, me hodological quali y emains
subop imal. This comp omises he quali y o s udies. The e o e, a g ea e e o is needed o enable he app op ia e
pe o mance o u u e s udies and inc ease he obus ness o e idence-based medicine in his impo an ield.
Keywo ds: Pallia i e ca e, Me hodological quali y, Risk o bias, Clinical ials
Backg ound
F om he i s ime i was used, he concep o “pallia i e
ca e”(PC) has su e ed a se ies o ans o ma ions in how
i is de ined and consequen ly in he ele an a ea o ope -
a ion and objec i es [1, 2]. In 2002 he Wo ld Heal h
O ganiza ion a i med ha PC imp o es he quali y o li e
o pa ien s and hei amilies acing p oblems associa ed
wi h li e- h ea ening illness, h ough he p e en ion and
elie o su e ing by means o ea ly iden i ica ion and im-
peccable assessmen and ea men o pain and o he
physical, psychosocial, and spi i ual issues [1, 3].
Changes in demog aphic ends, including he ageing
o popula ions and he inc eased li e expec ancy o
indi iduals wi h li e-limi ing illnesses, ha e inc eased
demand o high quali y PC se ices. Today, he ini i-
a ion o a ea men on he basis on wha is belie ed o
be e ec i e is no longe conside ed good clinical p ac-
ice [4]. A clinician in addi ion o his clinical expe ise,
mus ha e access o he bes a ailable e idence, should
ca e ully app aise i s quali y and assess i s applicabili y
o each indi idual pa ien [5, 6].
* Co espondence: [email p o ec ed]
1
Clinical Pha macology Uni , Ins i u o de Medicina Molecula , Facul y o
Medicine, Uni e si y o Lisbon, A enue P o esso Egas Moniz, 1649-028
Lisbon, Po ugal
3
Labo a o y o Clinical Pha macology and The apeu ics, Facul y o Medicine,
Uni e si y o Lisbon, A enue P o esso Egas Moniz, 1649-028 Lisbon, Po ugal
Full lis o au ho in o ma ion is a ailable a he end o he a icle
© The Au ho (s). 2017 Open Access This a icle is dis ibu ed unde he e ms o he C ea i e Commons A ibu ion 4.0
In e na ional License (h p://c ea i ecommons.o g/licenses/by/4.0/), which pe mi s un es ic ed use, dis ibu ion, and
ep oduc ion in any medium, p o ided you gi e app op ia e c edi o he o iginal au ho (s) and he sou ce, p o ide a link o
he C ea i e Commons license, and indica e i changes we e made. The C ea i e Commons Public Domain Dedica ion wai e
(h p://c ea i ecommons.o g/publicdomain/ze o/1.0/) applies o he da a made a ailable in his a icle, unless o he wise s a ed.
Bouça-Machado e al. BMC Pallia i e Ca e (2017) 16:10
DOI 10.1186/s12904-016-0181-9
Acco ding o a MEDLINE sea ch, he numbe o PC
clinical ials (CT) published has quad upled, om incep-
ion o 2005 [7, 8]. Whils his may be bene icial, ques ions
exis a ound he ype and quali y o he esea ch being
unde aken. P e ious e iews ha e concluded ha PC
s udies we e la gely desc ip i e, wi h a wide a ia ion in
sample size, in demog aphic and clinical aspec s and wi h
a lack o use o ecognised s anda d measu es and consid-
e a ion o key ou comes [5, 6, 9–12]. Visse e al. [5] s ud-
ied he eali y o e idence-based p ac ice in pallia i e ca e
and highligh ed addi ional p oblems like unpowe ed s ud-
ies, ec ui men di icul ies and high a i ion a es, inad-
equa e du a ion o ollow-up and di icul y in de ining
ou comes and a oiding pe o mance bias [5, 9, 13].
In esponse o he high a iabili y o clinical p ac ice
and he inc easing cos s and complexi y o ca e, e idence
is needed o de ine wha a e he mos e ec i e ea men s.
Good quali y andomized con olled ials (RCTs) a e he
gold s anda d o e alua ing he e icacy and e ec i eness
o heal h ca e in e en ions [14, 15]. Since p e ious publi-
ca ions showed a low numbe o andomized clinical ials
(RCT) in he pallia i e ca e ield, o achie e a mo e com-
p ehensi e iew o he apeu ic pallia i e ca e esea ch, we
designed a b oad sea ch s a egy including all ypes o
con olled clinical ials (CCT), o which RCT ep esen a
subg oup [14, 15]. The goal o his sys ema ic e iew was
o e alua e he me hodological cha ac e is ics o CCT in
pallia i e ca e and o assess hei quali y o epo ing.
Me hods
Li e a u e sea ch
We pe o med a MEDLINE sea ch h ough O id om in-
cep ion (1946) o Feb ua y 2015 using a p e-de ined sea ch
s a egy (Addi ional ile 1) designed by he au ho s based
on The Coch ane Collabo a ion’s highly sensi i e sea ch
s a egy o iden i y RCTs in he ield o pallia i e ca e.
S udy selec ion
Inclusion c i e ia o s udies we e:
p ospec i e con olled clinical s udy;
pha macological and non-pha macological
in e en ions;
s udies e alua ing pallia i e ca e in e en ions
(acco ding o each o he au ho s’de ini ion)
conduc ed in pa ien s and/o amily membe s o
ca egi e s, ega dless he place o ca e;
 ull-leng h a icle a ailable.
We excluded:
non-expe imen al s udies (obse a ional s udies,
sys ema ic e iews, me hodological s udies, s udy
p o ocols);
expe imen al s udies which did no e alua e
pallia i e ca e in e en ions;
expe imen al s udies e alua ing pallia i e ca e
in e en ions no di ec ly ocused in pa ien - amily
dyad (cos -e ec i eness analysis, e alua ion o pallia-
i e ca e se ices/uni s, and in e en ions di ec ed a
heal h p o essionals).
Ti les and abs ac s o ci a ions we e independen ly
p e-sc eened by wo e iewe s (RB, MR) acco ding o e-
iew s udy selec ion c i e ia. The inclusion o exclusion
c i e ia we e applied and s udies we e selec ed o con-
side a ion on he basis o ull ex epo s. Two e-
iewe s independen ly assessed he ull s udy epo s;
disag eemen s we e esol ed by consensus o by consul -
a ion wi h a hi d e iewe (JJF).
Da a ex ac ion and quali y assessmen
Be o e s udy selec ion, a da a ex ac ion o m wi h 43
i ems was de eloped, based on he checklis o guidelines
o he design and e alua ion o clinical ials (CONSORT,
SPIRIT) [16–18]. Da a ex ac ion was done manually by
wo esea che s (RBM, MR) wi hou any ex ac ion so -
wa e. Fi e domains we e analysed:
gene al in o ma ion ( i le o he CCT, name and
coun y o he co esponding au ho , language o
publica ion, yea and jou nal o publica ion, jou nal
impac ac o , a ea and ype o in e en ion, pe sonal
dimension and key poin s o p ac ice o PC
e alua ed, e hical app o al and in o med consen );
me hods (eligible c i e ia, ype o s udy design,
me hod o andomisa ion, achie emen o alloca ion
concealmen , ype o blinding, and du a ion o
ollow-up);
sample (in e en ion, o al numbe o andomised
pa ien s and numbe o pa ien s in each g oup,
du a ion and iming o ea men , d opou a e, and
sample size calcula ion);
da a analysis ( ype o analysis, s a is ical me hods
used, p e-de ined ou comes, assessmen ools, and
g oup compa abili y);
 esul s.
Included a icles we e classi ied by clinical domain (e.g.
oncology, neu ology) and ype o in e en ion. Fou ypes
o in e en ions we e conside ed: pha macological, non-
pha macological (all non-pha macological in e en ions
p o ided by heal h ca e p o essionals ha a e speci ically
men ioned as pa o he in e disciplina y pallia i e ca e
in e en ions [19]), non-pha macological complemen a y
he apies (all non-pha macological in e en ions, such as
musical and a oma he apy, ha a e no conside ed as pa
o he co e pallia i e ca e in e disciplina y in e en ions
Bouça-Machado e al. BMC Pallia i e Ca e (2017) 16:10 Page 2 o 12
[19]), and home-ca e based (all pha macological and non-
pha macological in e en ions p o ided in pa ien ’s
home).
We iden i ied PC miles ones ( ocus on whole-pe son,
pa ien and amily empowe men , good communica ion,
imp o emen o quali y o li e and eamwo k) mos ele-
an in he aims o each s udy. Based on hem, we p o-
ceeded wi h wo di e en ypes o classi ica ions, one
acco ding o he main pe sonal dimensions (physical,
psychological, social o spi i ual dimensions), and a sec-
ond le el in line wi h o he ac o s o PC p ac ice (com-
munica ion, symp oms con ol, amily suppo and eam
wo k) [20, 21].
The me hodological quali y o he included s udies was
assessed using he Coch ane Risk o Bias (RoB) ool [22].
This ool quan i ies he associa ion be ween ce ain design
ea u es and es ima es o ea men e ec s. The RoB ool
is a wo-pa ins umen and includes he ollowing a eas:
sequence gene a ion, alloca ion concealmen , blinding (o
pa icipan s, in es iga o s and ou come assessmen ), in-
comple e ou come da a, selec i e ou come epo ing and
“o he issues”. The i s pa e e s o he desc ip ion o
wha was epo ed in he ial, de ailed enough o a
judgemen o be made based on his in o ma ion. The sec-
ond pa app aises he isk o bias o each analysed a ea
and classi ies hem in h ee ca ego ies: low, high o un-
clea isk o bias [15, 23].
Independen ly, wo au ho s (RBM, MR) ex ac ed in-
o ma ion on indi idual i ems om all included s udies
and assessed he wo pa s in each s udy. Disc epancies
we e esol ed h ough discussion o by consul a ion
wi h a hi d e iewe (JJF).
S a is ical analysis
We summa ised he publica ion cha ac e is ics using
equencies and pe cen ages. Pooled odd a ios (OR) and
he 95% con idence in e al (CI) we e calcula ed using a
andom e ec s model. This me hod o e s summa y es i-
ma es by combining he indi idual esul s published by
independen esea che s. I inc eases powe and p o-
duces mo e p ecise summa y es ima es o he isk o
d opou be ween in e en ions and con ol g oups [24].
Di e ing d opou a es be ween ea men and con ol
a ms, wi h ewe pa ien s being ollowed up in one a m
han he o he , inc eases he isk o a i ion bias and he
possibili y o alse-nega i e esul s [22, 25]. Fo his ana-
lysis we used Re iew Manage 5.3.0 so wa e [22], Man el-
Haenzel me hod o accoun o he he e ogenei y (clinical
and me hodological) among s udies.
Resul s
The elec onic sea ch iden i ied 939 ci a ions. A e
sc eening abs ac s 120 a icles we e deemed po en ially
eligible. The applica ion o inclusion c i e ia excluded 13
s udies. The main easons o exclusion we e: epea ed
in he lis o e e ences (n= 3), duplica ed publica ions
(n= 8) and non-English language (n= 2) (Fig. 1).
Gene al ea u es
O he 107 clinical ials included (Addi ional ile 2),
12.2% (n= 13) we e published be ween 1989 and 1999,
45.8% (n= 49) be ween 2000 and 2009, and 41.1% (n=44)
be ween 2010 and 2015 (Fig. 2). S udies we e published in
i y-se en di e en jou nals, wi h he mos epo ed
being: Jou nal o Pain and Symp om Managemen
(14.9%, n= 16, impac ac o [IF]: 2.47), Pallia i e
Medicine (13.1%, n= 14, IF: 2.85), Jou nal o Pallia i e
Medicine (9.3%, n= 10, IF: 2.06) and Jou nal o Clinical
Oncology (5.6%, n= 6, IF: 17.9). Mos s udies we e con-
duc ed in he Uni ed S a es (USA) (26.2%, n=28), he
Uni ed Kingdom (UK) (21.5%, n= 23), Aus alia (11.2%,
n= 12), and Canada (6.5%, n= 7). Fi een pe cen (n=16)
o all he s udies lacked men ion o app o al by an e hics
commi ee.
Types o design
Eigh y- wo poin h ee pe cen (n= 88) o all he s udies
had a pa allel design and 17.7% (n= 19) had a c osso e
design.
The mos used compa a o was non-in e en ion
(con ol g oup pa icipan s did no ecei e any in e en-
ion o he du a ion o he s udy ollow-up)/bes sup-
po i e ca e (46.7%, n= 50) ollowed by placebo (27.1%,
n= 29) and o he in e en ions (25.2%, n= 27). The ana-
lysis o ype o in e en ion and ype o compa a o
demons a ed ha non-in e en ion/bes suppo i e ca e
was essen ially used in non-pha macological in e en-
ions (80%, n= 40), while o he in e en ions and pla-
cebo we e mo e used in pha macological in e en ions
(88.9%, n= 24 and 62.1%, n= 18). Ano he in e en ion
was chosen mo e o en han placebo in pha macological
in e en ions.
Follow-up du a ion a ied be ween s udies. The mos
common pe iods we e 1 mon h (14%, n= 15), 2 mon hs
and 2 weeks (9.3%, n= 10 each). The sho es ollow-up
was 30 min (a he end o an in e en ion) and
54 mon hs was he longes pe iod epo ed.
Eligibili y
Eligibili y c i e ia a ied signi ican ly h oughou s udies.
In he included s udies all pa ien s we e a leas 18 yea s
old and no s udies indica ed he gende o e hnici y o
pa icipan s. Acco ding o wha has been p e iously e-
po ed, oncological disease is o en an inclusion c i e -
ion. In h ee s udies (2.8%), demen ia was also an
inclusion c i e ion, while i was an exclusion c i e ion in
29 s udies (27.1%). The expec ed emaining li espan o
pa icipan s a ied be ween “less han a week o li e”and
Bouça-Machado e al. BMC Pallia i e Ca e (2017) 16:10 Page 3 o 12
24 mon hs, wi h 6 mon hs o li e being he mos com-
monly conside ed pe iod. In 66.4% (n= 71) o a icles
his da a was unknown.
Clinical domains
The h ee clinical domains mo e p esen in he pallia i e
ca e included s udies we e: oncology (56.1%, n= 60),
men al heal h (15.9%, n= 17) and gene al p ac ice (9.3%,
n= 10) (Fig. 3).
Types o in e en ions
Rega ding he ype o in e en ion, 44.9% (n= 48) o s ud-
ies e iewed we e non-pha macological in e en ions,
43.9% (n= 47) pha macological in e en ions, 7.5% (n=8)
non-pha macological complemen a y he apy in e en ions,
and 3.7% (n= 4) home-ca e based in e en ions (all
pha macological and non-pha macological in e en ions
p o ided in pa ien ’shome.SeeFig.4).
Pallia i e ca e classi ica ions
Wi h espec o he pe sonal dimension s udied, 63.6%
(n= 68) analysed he physical dimension, 13.1% (n= 14)
he psychological dimension, 14% (n= 15) he social di-
mension and 9.3% (n= 10) he spi i ual dimension. By
classi ying he s udies acco ding o he o he key poin s
o pallia i e ca e p ac ice we ound ha 70.1% (n= 75)
o he s udies we e based on symp om con ol e alu-
a ion, eamwo k and communica ion bo h ep esen ed
12.1% (n= 13) o s udies, and 5.6% (n= 6) s udies
highligh ed amily suppo .
Fig. 1 Flow diag am o s udy selec ion p ocess
Fig. 2 Numbe o clinical ials published o e ime
Bouça-Machado e al. BMC Pallia i e Ca e (2017) 16:10 Page 4 o 12
Ou comes and assessmen ools
As expec ed due o he b oad scope o his e iew, he e
was a signi ican di e si y o e alua ed clinical ou comes.
Howe e , in 40.2% o he included ials (n= 43) no p i-
ma y ou come was de ined. When men ioned, he mos
ci ed p ima y ou comes we e: pain in ensi y (20.3%, n=
13), imp o emen in quali y o li e (12.5%, n= 8), im-
p o emen in dyspnoea (9.4%, n= 6), and su i al a e
(7.8%, n= 5). The mos common seconda y ou comes
we e: imp o emen in quali y o li e (29.9%, n= 32), im-
p o emen in dep ession and anxie y (19.6%, n= 21), use
o escue doses o pallia i e ca e se ices (15.9%, n= 17),
o p esence o side e ec s (15%, n= 16). In he absence
o a p e-speci ied main ou come, we conside ed all ou -
comes as seconda y.
Fo ou come assessmen 137 di e en scales and
ques ionnai es we e used, wi h only ele en (8%) used
in mo e han i e s udies. Twen y (14.6%) o he 137
a e ecommended by he Na ional Pallia i e Ca e Re-
sea ch Cen e , 5 (3.7%) belong o he g oup o mos
used scales (Fig. 5).
S a is ic analysis
Fou s udies (3.7%) ailed o desc ibe s a is ical planning,
only one (0.9%) used desc ip i e analysis. In he majo i y
o s udies he analysis pe p o ocol was deduced om
he p esence o d opou s and he absence o in en ion-
o- ea analysis epo ing. Hal he s udies (50.5%, n=
54) used in en ion- o- ea analysis, 47.7% (n= 51) ana-
lysis pe p o ocol and in wo a icles (1.9%) i was no
Fig. 3 Dis ibu ion o included CTs ac oss medical ields
Fig. 4 Dis ibu ion o included CTs based on ypes o in e en ion
Bouça-Machado e al. BMC Pallia i e Ca e (2017) 16:10 Page 5 o 12

possible o conclude which s a is ical analysis had been
used. Sample size calcula ion was no indica ed in 38.3%
(n= 41) o s udies, and 8.4% (n= 9) used a con enience
sample. O he 57 s udies ha p esen ed a sample size
calcula ion, in only 36.8% (n= 21) was he numbe o in-
cluded pa ien s abo e he es ima ed sample size.
D opou s
The mean sample size was 113.1 (SD 139.1) [ ange
9–820] pa icipan s, wi h a median o 64.5. The mean
d opou equency (n= 99 s udies) was 22%, 40.2% o
he s udies had a d opou a e > 20% (cu -o used o
assess isk o bias). The main causes o a i ion we e symp-
om bu den and clinical de e io a ion. The clinical domains
and ypes o in e en ion wi h a highe pe cen age o s ud-
ies wi h a d opou a e > 20% we e: oncology (n= 30, 28%),
men al heal h (n= 11, 10.3%), pha macological (n= 29,
27.1%), and non-pha macological in e en ions (n= 23,
21.5%).
Pooled esul s om s udies ha epo ed one o mo e
d opou s (n= 91) showed highe d opou a es among
he ac i e in e en ion g oups (OR 1.32; 95% CI 1.07,
1.62). Howe e , despi e he use o a andom e ec s
model, he high le el o he e ogenei y limi s he accu -
acy o he me a-analysis esul s (Fig. 6).
Quali y o epo ing analysis
Only in wo pape s (1.9%) we e all domains conside ed as
ha ing low RoB, while in 33 (30.8%) he e was a low RoB
in a leas hal o hem (4/7 domains). In eigh s udies
(7.5%) he e was a high RoB in a leas hal he ca -
ego ies and in 39 s udies (36.5%) he isk o bias was
unclea (Addi ional ile 3; Fig. 7).
The pe cen age o ials in he las 5 yea s ha had a
low RoB in a leas hal he domains was highe compa ed
wi h ials published ea lie (33.3% s 29.7%). Howe e ,
he pe cen age o s udies high o unclea RoB in a leas
hal he domains in he las 5 yea s was also highe (high
Rob –9.1% s 6.8%; unclea Rob –42.4% s 33.8%).
Only one s udy was no andomised. Compu e -
gene a ed andomisa ion was he mos used mechanism,
p esen in 37.4% o s udies (n= 40). Rega ding he ype
o andomisa ion: 23.4% (n= 25) used andomisa ion in
blocks, 15% (n= 16) s a i ied, 4.7% (n= 5) simple and
0.9% (n= 1) used a minimisa ion me hod. Mos s udies
used a pe son, unconnec ed wi h he s udy (e.g., an inde-
penden s a is ical colleague o he pha macis ), o gua -
an ee alloca ion concealmen .
Rega ding blinding (o pa icipan s, in es iga o s and
ou come assessmen ), 19.6% (n= 21) o s udies we e
double-blind, 14% (n= 15) single-blind, in 6.5% (n=7)
all elemen s we e blinded and 19.6% (n= 21) we e open-
label s udies. In 40.2% (n= 43) his in o ma ion was no
epo ed.
Acco ding o he ins uc ions o he Coch ane ool,
when he p ima y ou come was no explici , isk o bias
was conside ed o be high, since i was no clea i he
a iables we e chosen o no based on he esul s.
Discussion
This e iew iden i ied 107 CCTs assessing PC in e en-
ions o pa ien s and/o amilies, he majo i y o hem
pe o med in he USA and he UK. Only one s udy was
no andomized. The amoun o missing da a is e y
high in almos all me hodological ac o s e alua ed.
O e all he e is no da a om he ial quali y app aisal
o sugges ha epo ing o me hods is imp o ing.
De ining “pallia i e ca e”: who, wha in e en ions, when?
In ou e iew, we ha e chosen o accep as a pallia i e
in e en ion ha which he au ho s assumed o be such.
As men ioned be o e, wi h inc eased awa eness ha
Fig. 5 Dis ibu ion o he mos used e alua ion scales in included s udies
Bouça-Machado e al. BMC Pallia i e Ca e (2017) 16:10 Page 6 o 12
non-oncology pa ien s could bene i om a pallia i e ap-
p oach, a se ies o ans o ma ions in he concep , in e -
en ions and objec i es o “pallia i e ca e”occu ed [1].
This di e si y is e lec ed in he lack o a common lexi-
con in PC co e e ms (such as “pallia i e ca e”o “end-
o -li e”) making i no only di icul o ensu e ha all
Fig. 6 Fo es plo compa ing d opou s be ween in e en ion and con ol g oup
Bouça-Machado e al. BMC Pallia i e Ca e (2017) 16:10 Page 7 o 12
eade s acing he same s udy each simila conclusions,
bu also o de ine he popula ion and speci ic in e en-
ions o pallia i e ca e [14, 26, 27]. O he e iews, such
as Lo enz e al. in 2005 e iew ha in ended o e alua e
he e idence in he ield om he pe spec i e o con-
ce ns impo an o pa ien s, ca egi e s, and he heal h
ca e sys em epo ed he same di icul ies, and in 2003
Bausewein e al., in a e iew on he challenges de ining
PC, highligh ed he lack o cla i y in de ini ion and e -
minology ega ding his subjec [28, 29].
Who?
Ou esul s showed ha he e was a clea p edominance
o in e en ions di ec ed o oncological pa ien s (e.g. he
compa ison o wo di e en me hods o he apy adminis-
a ion in he ea men o b eak h ough pain in pa ien s
wi h cance ) co esponding o 56.1% o he included s ud-
ies. I a i s his looks no mal due o he ini ial ocus o
PC on pa ien s dying om cance , he di e ence be ween
he pe cen age o oncology s udies and s udies o o he
special ies (56.1% s. 43.9%) seems o show ha we a e
now beginning o ge used o he idea o PC in cance , bu
o o he diseases his is a om eali y. I is also ele an
ha in 27.1% o he s udies demen ia was an exclusion c i-
e ion. Wi h people li ing longe and su e ing mo e om
diseases ha a e associa ed wi h cogni i e impai men , he
numbe o people who a e demen ed and may bene i
om PC in e en ion is inc easing. The e o e, cogni i e
impai men and demen ia should no be excluded om
he pallia i e ca e popula ion, since his can h ea en he
ex e nal alidi y o s udies [30, 31].
Wha in e en ions?
Rega ding he ype o in e en ion, he numbe o s udies
assessing pha macological and non-pha macological in e -
en ions was e y simila (43.9% s. 44.9%), wi h he ma-
jo i y o hem e alua ing in e en ions o symp oma ic
con ol (70.1%). O he ypes o in e en ions, such as
non-pha macologic complemen a y he apies (7.5%) and
home-ca e based (3.7%), o di e en aspec s o ca e such
as communica ion (12.1%) o amily suppo (5.6%) we e
less co e ed. Albe s e al. [26] and Hui e al. [32] in wo
sys ema ic e iews on me hodological issues in PC,
poin ed ou he imbalance be ween pha macological in e -
en ions and o he in e en ions no ela ed wi h symp-
om con ol, which ep esen ed 5% o less o he o al
RCTs in pallia i e ca e. The Na ional Ins i u e o Heal h
in he USA highligh ed ha ew publica ions on pallia i e
ca e esea ch e lec ed he g owing needs o pa ien s [33].
E en in he con ex o symp oma ic con ol esea ch, im-
po an gaps in clinical e idence should be add essed. Fo
example nonpain symp oms, such as b ea hlessness o de-
li ium, a e s ill poo ly unde s ood and symp om bu den
con inues o be he main complain o pa ien s and cause
o d opou om s udies despi e elie o dis essing symp-
oms being conside ed one o he guiding p inciples o
pallia i e ca e p ac i ione s [10, 34].
When?
In his e iew he p ognosis o pa ien s anged be ween
“less han one week” o 24 mon hs. The de ini ion o
pallia i e ca e poin s owa ds a popula ion wi h li e-
limi ing disease and, when cu e is no possible, wha is
equen ly unde s ood as he ca e o pa ien s in hei las
weeks o days o li e [10, 35]. Howe e , some diseases,
especially ch onic diseases, se e ely a ec he quali y o
li e o pa ien s and amily membe s o many yea s, his
led o conside ing pallia i e ca e ea lie , and including in
mo e ecen de ini ions he ini ia ion o pallia i e ca e a
he ime o diagnosis and p o ided conco dan ly wi h all
o he disease-di ec ed o cu a i e ea men s [10].
The place o RCTs in pallia i e ca e esea ch
O he 939 iden i ied ci a ions, only 11.4% (n= 107) we e
CCTs e alua ing pallia i e ca e in e en ions in pa ien s
and/o amilies. This is in line wi h p e ious me hodo-
logical e iews, Hui e al. [5, 32] epo ed in 2011 ha
RCTs only comp ised 5.6% (n= 47) o he s udies and in
Fig. 7 Risk o bias in included s udies assessed using he Coch ane ool
Bouça-Machado e al. BMC Pallia i e Ca e (2017) 16:10 Page 8 o 12
2006 Kaasa e al. [27], d ew a en ion o he ac ha
only 4.3% o publica ions we e p ospec i e e alua ions
o in e en ions wi h case se ies (50.7%, n= 462) and
c oss-sec ional s udies (17.7%, n= 149) being he mos
common s udy designs. In 2014, Aoun and Nekolaichuk
[9] epo ed ha Coch ane e iews in pallia i e ca e
ailed o p o ide good e idence because o he ew num-
be s and poo in e nal and ex e nal alidi y o p ima y
s udies. Al hough hese ypes o s udies a e a mino i y in
pallia i e ca e esea ch, ou esul s show ha he num-
be CCTs has inc eased in ecen yea s.
Challenges pe o ming RCTs in pallia i e ca e
Rec ui men , a i ion and powe ed samples
The samples o he included s udies a y be ween 9 and
820 pa icipan s, wi h a median o 64.5. I is no uncom-
mon o inds a emen ssugges ing ha i isune hical o
in ol e people wi h pallia i e ca e needs in esea ch
because o hei inc eased ulne abili y [27]. Aoun and co-
wo ke s [9], in a e iew abou he challenges o e idence-
based medicine (EBM) in pallia i e ca e esea ch demon-
s a ed p ecisely he opposi e: he pa icipa ion in expe i-
men al p o ocols was no pe cei ed as an addi ional
s ess, bu a he like a pe sonal gain in a sel less pe spec-
i e ela ed wi h a mode a e- o-high bene i . To ca e-
gi e s, his collabo a ion is seen as an added alue o
pa ien s, o hemsel es, and o u u e amilies ha need
pallia i e ca e assis ance. Recen ly, he esul o a wo k-
shop and consensus exe cise (MORECa e s udy), abou
bes p ac ice on e hical conce ns in PC esea ch [33],
a i med ha i is e hically desi able o pa ien s and hei
amilies wi h pallia i e ca e needs o be o e ed he oppo -
uni y o be in ol ed in esea ch and eminded o he
exis ence o ele an in e na ional ecommenda ions o
o e come some o he e hical challenges aced. Abe ne hy
e al. [36], in a e iew on key insigh s o enhance he en-
olmen in pallia i e ca e ials, sugges ed s a egies o
success ully ec ui pa ien s o la ge-scale andomised
clinical ials, o example whe e app op ia e he adop ion
o lexible in e en ions, he educ ion o ea men ime
pe iods, and he educ ion o he numbe o s udy assess-
men s including in pa icula hose ha a e in asi e o
ime consuming.
Ou esul s show a median a i ion a e o 22%. A e-
iew by Hui e al. [13] ound a median a i ion a e o
44% in pallia i e oncological CT. When using a cu -o
o ≤20% o losses o ollow-up and compa ing wi h a e-
iew o 71 RCT in ou op medical jou nals showed a
d opou a es o ≥20% in 18% o he ials [25], we can
assume ou 40.2% o s udies abo e his cu -o as a high
d opou a e. I is una oidable o ha e some missing da a,
bu igno ing i is no accep able, since i ep esen s a sig-
ni ican isk o he powe , p ecision and gene alizabili y o
ials esul s. Looking a ou pooled esul s, he e was a
highe pe cen age o d opou s in in e en ion a m. Hus-
sain e al. [37] in a e iew on missing da a in PC RCT e-
po ed a simila esul wi h a high d opou a e in
in e en ion a m. Howe e , as we men ioned in esul s
sec ion he e was a high le el o he e ogenei y be ween
s udies ha didn’ allow o be conclusi e in ela ion o his
ques ion. Fu he mo e, Bell e al. [25] sugges ed ha o
an accu a e analysis o a i ion bias in pooled esul s is
no enough o know he di e en ial d opou a es, is also
necessa y ake in o accoun he ype o missingness (a
andom, comple ely a andom o no a andom), he
analysis me hods and he e ec ha is being es ima ed.
The au ho s sugges ed he use o mixed models me hods
as a s a egy o es ima e unbiased ea men e ec s, unde
assump ions ega ding he misingness mechanism(s).
O he 107 included clinical ials, only 53.3% o s udies
epo ed a sample size calcula ion, and o hese only 36.8%
(n= 21) eached he minimum o pa ien s es ima ed. This
is a majo p oblem in clinical esea ch because, as men-
ioned abo e, i can be misleading ei he by missing ealis ic
mode a e ea men e ec s ha would be clinically impo -
an , o by o e es ima ing he size o a ea men e ec and
inding i s a is ically signi ican pu ely due o chance [38].
Visse e al. [5] al eady epo ed in 2015 ha mos o he
p ima y s udies used in pallia i e ca e e iews we e me h-
odologically lawed and hose ha we e conside ed highe
quali y we e inadequa ely powe ed.
Ou comes and assessmen ools
Besides he la ge di e si y o s udy ou comes, ou esul s
demons a e he absence in a signi ican pe cen age o
s udies (40.2%) o an explici de ined p ima y ou come,
which inc eases he isk o epo ing bias since i is no
ensu ed ha a iables p esen ed we e no chosen based
on he s udy esul s. Because o he g ea he e ogenei y
in popula ion and ype o in e en ions, he e is s ill a
lack o consensus in pallia i e ca e ield abou he bes
ou come measu es and clinically meaning ul di e ences
o each ou come.
In his e iew only 3.7% o he applied scales a e in he
lis ecommended by The Na ional Pallia i e Ca e Re-
sea ch Cen e . The choice o assessmen ools is e y
impo an in s udy p o ocols and one o he challenges
o eaching high quali y esea ch. Al hough se e al in-
s umen s can be used o assess ou comes, no all we e
de eloped and alida ed o use in a pallia i e ca e popu-
la ion and so no he mos app op ia e [39].
The use o placebo-con ol ials
Ou esul s show ha he mos used compa a o was
non-in e en ion/bes suppo i e ca e (46.7%, n= 50).
Bes suppo i e ca e (BSC) in e en ions we e de ined
by Jassem e al. (2008) as “ ea men adminis e ed wi h
he in en o maximize quali y o li e wi hou a speci ic
Bouça-Machado e al. BMC Pallia i e Ca e (2017) 16:10 Page 9 o 12