RESEARCH ARTICLE Open Access
Measu ing he e iciency o Pales inian
public hospi als du ing 2010–2015: an
applica ion o a wo-s age DEA me hod
Wasim I. M. Sul an
1,2*
and José C ispim
1
Abs ac
Backg ound: While heal h needs and expendi u e in he Occupied Pales inian Te i o ies (OPT) a e g owing, he
in e na ional dona ions a e declining and he economic si ua ion is wo sening. The pu pose o his pape is
wo old, o e alua e he p oduc i e e iciency o public hospi als in Wes Bank and o s udy con ex ual ac o s
con ibu ing o e iciency di e ences.
Me hods: This s udy examined echnical e iciency among 11 public hospi als in Wes Bank om 2010 h ough 2015
a ge ing a o al o 66 obse a ions. Na ionally ep esen a i e da a we e ex ac ed om he o icial annual heal h epo s.
We applied inpu -o ien ed Da a En elopmen Analysis (DEA) models o es ima e e iciency sco es. To elabo a e u he
on pe o mance, we used Tobi eg ession o iden i y con ex ual ac o s whose impac on ine icien pe o mance is
s a is ically signi ican .
Resul s: Despi e he inc ease in e iciency mean sco es by 4% om 2010 o 2015, indings show po en ial sa ings o
14.5% o esou ce consump ion wi hou educing he olume o he p o ided se ices. The signi ican Tobi model
showed ou p edic o s explaining he ine icien pe o mance o a hospi al (p< 0.01) a e: bed occupancy a e (BOR);
he ou pa ien -inpa ien a io (OPIPR); hospi al’s size (SIZE); and he a ailabili y o p ima y heal hca e cen e s wi hin he
hospi al’s ca chmen a ea (PRC). The e is a s ong e ec o OPIPR on e iciency di e ences be ween hospi als: A one
uni inc ease in OPIPR will lead a dec ease o 19.7% in he p edic ed ine iciency le el holding all o he ac o s cons an .
Conclusion: To da e, no p e ious s udies ha e examined he e iciency o public hospi als in he OPT. Ou wo k
iden i ied hei e iciency le els o po en ial imp o emen s and he de e minan s o e icien pe o mance. Based on
he measu emen o e iciency, he gene a ed in o ma ion may guide hospi als’manage s, policymake s, and
in e na ional dono s imp o ing he pe o mance o he main na ional heal hca e p o ide . The scope o his s udy is
limi ed o public hospi als in Wes Bank. Fo a be e unde s anding o he Pales inian ma ke , u he esea ch on
p i a e hospi als and hospi als in Gaza S ip will be use ul.
Keywo ds: Public hospi als, E iciency, 2-DEA, Tobi eg ession, Wes Bank
Backg ound
The heal hca e sys em in he Occupied Pales inian Te i o -
ies (OPT) is in luenced by he ambiguous poli ical en i on-
men wi hin which i is enac ing [1]. The OPT (Wes Bank,
Eas Je usalem, and Gaza S ip) is a coun y in ch onic con-
lic and economic eme gency [2]. The ne e -ending con lic
be ween he Pales inians and he Is aelis seemed o come o
an end when he Middle Eas peace p ocess was se led, pa -
icula ly, a e he Mad id con e ence in 1991, hen he Oslo
Acco ds in 1993 and he es ablishmen o he Pales inian
Au ho i y (PA) in 1994. Hence o h, building he capaci y
o he Pales inian public heal hca e sec o e ol ed [3], and
had unde gone se e al e o ms. Re o ms we e hea ily sub-
sidized by in e na ional dona ions [4], as e o s made by
he in e na ional communi y o esol e he con lic in
Pales ine-Is ael h ough economic encou agemen s [5].
Despi e he no iceable p og ess in ebuilding he ins i-
u ions o he ye o be “The S a e o Pales ine,”g ound
* Co espondence: [email p o ec ed]
1
School o Economics and Managemen , Uni e si y o Minho, 4710-057
B aga, Po ugal
2
P.O. Box 198, Heb on, Pales ine
© The Au ho (s). 2018 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.
Sul an and C ispim BMC Heal h Se ices Resea ch (2018) 18:381
h ps://doi.o g/10.1186/s12913-018-3228-1
eali y sugges s o he wise. The si ua ion is emaining
complica ed and p oblema ic as wi nessed by mo e isola-
ion and mo e es ic ions on mo emen be ween Wes
Bank (WB) and Gaza S ip (GS) and be ween ci ies wi hin
WB. The Pales inians a e no allowed o a el eely be-
ween he OPT egions [1]. To da e, he Is aelis con ol
o e wa e , elec ici y, bo de s, and anspo amongs
o he in as uc u al ma e s, while, he Pales inians ha e
limi ed con ol o e hei own a ai s. This unique con ex
has implica ions on he p io i y se ings and he p ocess o
heal h policy implemen a ion [6]. The e o e, in p ac ice,
he in eg a ion o heal h policies and heal h deli e y ope -
a ions is no jus a ma e o combining he wo.
The Pales inian Minis y o Heal h is he leading heal h-
ca e p o ide including hospi al ca e and bea s he mos
subs an ial bu den o mee he cons an g ow h in he de-
mand o heal hca e se ices. On a e age, heal h expend-
i u e eco ded consis en annual g ow h a e o 7%. The
o al heal h expendi u e inc eased om $400 million in
he yea 2000 o $1400 million in he yea 2015; he la e
accoun ed o 10.7% o he coun y’sG ossDomes ic
P oduc (GDP). The public eimbu semen schemes ep-
esen 62.5% o he o al heal h expendi u es [7].
Hospi als, wi h 6006 beds, a e he main heal hca e
p o ide s o se e 4.48 million people li ing in WB and
GS. Fo y- wo pe cen o he o al expendi u e is spen
on hospi al ca e (i.e., 4.5% o he coun y’s GDP) [8]. Be-
cause hospi als make up a la ge po ion o heal hca e ex-
pendi u e, hospi als a e a po en ially la ge sou ce o cos
sa ings. The e o e, he analysis o his s udy was
in ended o cap u e po en ial gains in he e iciency o
public hospi als ha may ha e a subs an ial con ibu ion
o la ge po en ial cos -sa ings o he coun y’s heal hca e
expendi u e [9–11].
Mo eo e , he applied go e nmen al heal h insu ance
scheme co e s mos o he Pales inians, by which hey
a e en i led o public se ices, had inc eased he bu den
on public hospi als. The e o e, public hospi als (61.1% o
all he hospi al beds) a e c owded and unc ioning a
high bed occupancy a es o e en o e occupied [2]. To
ca e o he inc easing heal h demand on heal hca e se -
ices, he Pales inian Minis y o Heal h (PMoH) allo-
ca es abou 40% o i s budge o pu chase hospi al
se ices om o he e e al hospi als wi hin he coun y
o ab oad such as hospi als in Jo dan [12]. Recen ly, he
Wo ld Heal h O ganiza ion (WHO) epo indica ed
ha he decline in dono s’suppo and he unique poli -
ical si ua ion o he Pales inians ha e se ious e ec s on
he scope and quali y o heal h condi ions [13].
The pu pose o his wo k is wo old, analyzing he e i-
ciency o he public hospi als in Wes Bank; and e alua ing
he en i onmen al ac o s a ec ing hei p oduc i i y. Keep-
ing in mind ha he hospi al echnical e iciency equi es
he use o minimum inpu o p oduce a gi en le el o ou pu
[10] and ha he abili y o a hospi al o ans o m inpu s
in o ou pu s is in luenced by i s manage ial e iciency as well
as he ex e nal ope a ing en i onmen [14,15].
The scope o ou wo k is limi ed o public hospi als in
Wes Bank. Hospi als in Gaza S ip a e excluded in his
wo k due o many limi a ions: (1) The geog aphical sep-
a a ion be ween Wes Bank and Gaza, he Pales inians
a e no allowed o a el ac oss hem; (2) The 2008 and
2014 wa s agains Gaza makes he con ex o hospi al
ope a ions incompa able; (3) The Pales inian in e nal
con lic since 2006 escala ed wi h he spli o Pales inian
Au ho i y in o one go e nmen in WB and ano he in
GS, hence, he ope a ional da a o hospi als in GS is un-
eliable. The e o e, he main scope o his pape s is o
examine he echnical e iciency o 11 public hospi als
ou o 13 public hospi als wo king in Wes Bank du ing
2010–2015 (i.e., 66 obse a ions).
We conduc ed seconda y esea ch o ind s udies
e alua ing he pe o mance o heal hca e p o ide s in
Pales ine; o da e, he e a e no p e ious s udies conce n-
ing he opic. The exis ing ele an li e a u e desc ibes
he ansi ional con ex and he complica ions wi hin he
coun y’s heal hca e sys em in Pales ine [2–4,6,16–19].
Hence, imp o ing pe o mance among he Pales inian
public hospi als by pe o mance measu emen is a s aigh -
o wa d need. The gene a ed in o ma ion will p o ide alu-
able insigh s o hospi al manage s who make ope a ional
decisions and o policymake s and in e na ional dono s
who may in luence he ex e nal ope a ing en i onmen by
egula ions, subsidies o by o he policy measu es.
Da a En elopmen Analysis (DEA) is a uni e sal me h-
odology in heal hca e e alua ion and widely used
non-pa ame ic me hodology o e alua e pe o mance
[20–22]. Since he ad en o DEA by Cha nes e al. [23],
mo e han 10 housand s udies had been published which
es ima ed he pe o mance o di e en kinds o en i ies
and p oduc ion ac i i ies including he heal hca e sec o
[24]. Recen DEA s udies ex end he analysis o in es iga e
a ia ions in hospi al pe o mance o e yea s and o iden-
i y con ex ual d i e s o e icien p ac ices [25].
Due o he lack o da a in de eloping coun ies, ew
empi ical wo ks applied he da a-based me hodology o
DEA models [26]. To da e, no s udies ha e examined
he pe o mance o public hospi als in Pales ine o po-
en ial imp o emen s. The e o e, his wo k add esses a
DEA li e a u e gap by analyzing he e iciency o he
public hospi als and iden i ying con ex ual d i e s o in-
e icien pe o mance in a de eloping coun y, namely,
Pales ine. In esponse o his need, ou endea o goes o
achie e he ollowing esea ch objec i es: (1) e alua e
how Pales inian public hospi als u ilize esou ces while
ca ing o hei pa ien s om 2010 o 2015; and (2) ex-
plo e en i onmen al e ec s associa ed wi h he e icien
use o hospi al esou ces.
Sul an and C ispim BMC Heal h Se ices Resea ch (2018) 18:381 Page 2 o 17
We apply wo-s age da a en elopmen analysis (2-DEA)
whe e he e icien on ie and he hospi al le el e iciency
sco e a e es ima ed wi h DEA model in he i s s age, and
he e iciency es ima es a e eg essed on con ex ual ac o s
in he second s age [15,27]. In s age 1, we calcula e he e -
iciency wi h which physical inpu s p oduce ou pu . In
s age 2, we apply Tobi eg ession which is commonly
used o ela e e iciency sco es o ac o s expec ed o in lu-
ence e iciency while hese ac o s a e no unde he con-
ol o hospi al manage s [14,28,29].
Empi ical con ex
The whole a ea o he OPTs is 6170 km
2
o which
5800 km
2
is he a ea o WB, and 365 km
2
is he a ea o
GS. The Pales inian heal hca e sys em comp ises i e main
p o ide s o heal hca e se ices: (1) The Pales inian Minis-
y o Heal h (PMoH) and ep esen s he public sec o ; his
sec o comp ises p ima y heal hca e cen e s and public
hospi als. These hospi als a e owned and adminis e ed by
he Pales inian Minis y o Heal h. They a e gene al hospi-
als ha p o ide p ima y and seconda y heal hca e se ices,
howe e ; no public hospi al p o ides e ia y se ices. (2)
The Uni ed Na ions Relie and Wo ks Agency o Pales ine
Re ugees (UNRWA); (3) Non-Go e nmen al O ganiza ions
(NGOs); (4) Pales inian Mili a y Medical Se ices (PMMS);
and (5) P i a e o -p o i o ganiza ions. Acco ding o he
Pales inian Cen al Bu eau o S a is ics (PCBS), hese p o-
ide s manage80 hospi als wi h a capaci y o 6006 hospi al
beds o se e 4.88 million people li ing in OPT, o which
2.97 million a e li ing in WB, and 1.91 million a e li ing in
GS. The median age o he Pales inians is 19.8 yea s, and
39.4% o he popula ion is unde 15 yea s old. The age
g oup (0–4 yea s) is 15% while o he age g oup o e
65 yea s cons i u e only 2.9% o he popula ion [12,30].
The e a e 50 hospi als a e ope a ing in WB including Eas
Je usalem (60.1% o o al beds), and 30 hospi als a e ope a -
ing in GS (39.9% o all he beds). 73% o all he hospi al beds
a e gene al beds, 19% a e specialized beds, 3.1% ehabili a-
ion a e beds and 4.9% a e ma e ni y beds (Table 1).
In Wes Bank, public hospi als a e dis ibu ed in 11
adminis a i e a eas (go e no a es). They a e Jenin,
Tubas, Tulka m, Nablus, Qalqillya, Sal i , Ramallah, Je i-
cho, Be hlehem, Heb on, and Eas Je usalem. Howe e ,
due o poli ical easons, he e is no Pales inian public
hospi al in Eas Je usalem. The e o e, he included hos-
pi als in his s udy a e 13 public hospi als wi h a capaci y
o 1594 beds wo king in WB. To ha e a homogeneous
sample o gene al hospi als, we excluded wo hospi als
om he analysis: a new hospi al wi h 37 beds was es ab-
lished in 2014 (P12 in Table 2), da a is no a ailable om
2010 o 2013; he o he hospi al is psychia ic wi h 180
beds. As a esul , we analyze he e iciency o 11 public
hospi als (P01-P11 in Table 2) om 2010 o 2015 (66
obse a ions). Table 2illus a es he sample cha ac e is-
ics and ele an ma ke a ibu es du ing 2015.
P oduc ion model and a iables
The abili y o a hospi al o ans o m inpu s in o ou pu s
is in luenced by i s manage ial e iciency (p ac ices) and
ex e nal ope a ing en i onmen (ope a ional condi ions)
[15]. The e o e, ela ing he measu es o ine iciency o
he su ounding con ex ual ac o s p o ides a be e un-
de s anding o e iciency di e ences and de e mines he
key pe o mance d i e s ac oss hospi als [31]. The OPT
has a agmen ed landscape o heal hca e p o ide s in-
cluding hospi als which e ol ed ac oss di e en egimes
[1,4]. Howe e , he geopoli ical se ing o he OPT
poses challenges o heal hca e deli e y and access, he e-
o e, i is belie ed ha en i onmen al ac o s ouch he
p oduc ion o heal hca e se ices and should be in-
cluded in ou analysis. Figu e 1shows he ela ionships
be ween inpu -ou pu measu es and con ex ual ac o s.
Di e en inpu and ou pu se s had been used in he DEA
li e a u e o analyze he e iciency o hospi als [32,33]. The
basic p inciple, o iden i y a iables, is o ha e a clea unde -
s anding o he “p ocess”being e alua ed among pee hospi-
als [34]. The in es iga ed hospi als a e all gene al hospi als;
hey a e designed o p o ide p ima y and seconda y heal h
se ices, hey don’ p o ide e ia y heal h se ices. The e-
o e, we included inpu -ou pu measu es ha make a p ac-
ical sense o he Pales inian public hospi al se ings. We
used ou pu measu es ha ep esen he le el o public
heal h bene i s achie ed in espec o h ee unc ional a eas;
admissions, ou pa ien isi s, and eme gency se ices. Since
he o he ac i i ies wi hin he hospi al (e.g., labo a o y es s,
deli e ies, su gical ope a ions, adiology ac i i ies) a e highly
co ela ed wi h he h ee measu es, we did no include hem
in he se o ou pu s [35].
We included h ee ou pu measu es, hey a e: (1) in-
pa ien se ices as measu ed by he o al numbe o an-
nual ca e days a he han a numbe o cases o accoun
o case-mix adjus men [36]; (2) ou pa ien se ices as
measu ed by he o al numbe o annual isi s [33]; and
(3) he eme gency se ices as measu ed by he o al an-
nual numbe o cases se ed wi hou admission [37].
Inpa ien days ep esen he o al annual du a ion o
pa ien admissions and he u iliza ion o clinical and
nonclinical inpu s, such as nu sing ca e, pha maceu ical
i ems, pa amedical suppo se ices, and adminis a i e
se ices. Ou pa ien isi s ep esen he u iliza ion o he
ou pa ien clinics and he dedica ed clinical and admin-
is a i e esou ces o hese clinics. In Pales ine, he
eme gency depa men s and he ambulance se ices a e
i al ou pu s and ep esen he u iliza ion o a conside -
able amoun o esou ces in he public hospi als. The ea-
sons behind he impe a i e ole o eme gency se ices a e:
(1) he hospi al eme gency depa men s become he i s
Sul an and C ispim BMC Heal h Se ices Resea ch (2018) 18:381 Page 3 o 17
choice o pa ien s seeking ea men because amily p ac ice
model is absen in Pales ine; (2) he p ima y heal hca e cen-
e s wo k only o 6 hou s a day, and 5 days a week, hey
p o ide a mino ole o eme gency and ambulance se ices;
and (3) he majo i y o he popula ion is co e ed by he go -
e nmen heal h insu ance scheme by which hey a e en i led
o he eme gency depa men s in public hospi als [2].
In line wi h o he DEA li e a u e [38–40], we included
ou inpu measu es. They cha ac e ize he employed
labo and capi al. Labou inpu measu es comp ise h ee
g oups o pe sonnel, he doc o s, he heal hca e
ull- ime employees FTEs (e.g., Nu ses, echnicians, and
o he employees in pa a-medical depa men s), and he
adminis a i e FTEs [41]. Capi al inpu measu e was
ep esen ed by he numbe o hospi al beds [42]. Da a
on o he esou ces, such as d ugs, labo a o y es s, o in-
s umen s we e no a ailable o he included hospi als.
As o he impac o he en i onmen on he p oduc -
i i y o he public hospi als in Wes Bank, we conside ed
en ac o s (Table 3). These ac o s a e o ganized in o
Table 1 Dis ibu ion o hospi al beds and p ima y heal hca e cen e s in OPTs in 2016
Hospi aliza ion
Type o hospi aliza ion Regions Public
a
O he s
b
By egion Hospi al beds
Gene al WB 1414 (32.2%) 1222 (28%) 2636 (60.2%)
GS 1328 (30.2%) 421 (9.6%) 1749 (39.8%) 4385 (73%)
Specialized WB 180 (15.7%) 437 (38.2%) 617 (53.9%)
GS 293 (25.6%) 234 (20.5%) 527 (46.1%) 1144 (19%)
Rehabili a ion WB 0.0 141 (76.2%) 141 (76.2%)
GS 0.0 44 (23.8%) 44 (23.8%) 185 (3.1%)
Ma e ni y WB 0.0 213 (72.9%) 213 (72.9%)
GS 43 (14.7%) 36 (12.4%) 79 (27.1%) 292 (4.9%)
To al by egion WB 1594 (44.2%) 2013 (55.8%) 3607 (60%)
GS 1664 (69.4%) 735 (30.6%) 2399 (40%)
To al beds 3258 (54.3%) 2748 (45.7%) 6006 (100%) 6006 (100%)
Public P ima y Ca e
Cen e s (PHCs)
WB 422 (69.4%) 186 (30.6%) 608 (80.0%)
GS 49 (32.2%) 103 (67.8%) 152 (20.0%)
To al PHC 471 (62.0%) 289 (38.0%) 760 (100%)
Public
a
, hospi als o PHCs a e owned and adminis e ed by he Pales inian Minis y o Heal h
O he s
b
, hospi als o PHCs a e no owned no adminis e ed by he Pales inian Minis y o Heal h
Table 2 Selec ed cha ac e is ics o he ope a ing public hospi als in Wes Bank (2015)
Go e no a e Ma ke cha ac e is ics Public hospi al cha ac e is ics
PHC/10000 Beds/10000 % public beds Public hospi als Beds Occupancy a e Hosp.
Heb on 2.22 9.0 48.9 Abu al Hasan 36 101.3 P01
Sal i 4.08 7.1 100.0 Yasse A a a 50 71.9 P02
Je icho 3.85 10.4 56.4 Je icho 54 71.7 P03
Nablus 1.84 16.9 40.8 Wa ani 55 86.0 P04
Qalqilya 3.51 10.9 47.9 D. Nazal 58 95.0 P05
Tulka m 2.36 9.3 69.1 Thabi Thabi 117 71.5 P06
Be hlehem 2.04 27.3 22.2 Al Hussein 131 79.4 P07
Jenin 2.09 7.1 73.8 Khaleel S. 163 90.1 P08
Nablus 1.84 16.9 40.8 Ra edia 200 87.6 P09
Ramallah 2.18 12.2 56.1 Med. Complex 238 97.6 P10
Heb on 2.22 9.0 48.9 Alia 275 120.4 P11
Tubas –5.7 –The Tu kish 37 63.1 P12
a
PHC P ima y Heal h Ca e Cen e s
a
Hospi al P12 is excluded, a ailable da a is limi ed o 2014 and 2015
Sul an and C ispim BMC Heal h Se ices Resea ch (2018) 18:381 Page 4 o 17
h ee se s: (1) Fac o s had been p e iously s udied by
o he esea che s such as he bed occupancy a e (BOR)
and he a e age leng h o s ay (ALOS). (2) Fac o s ep e-
sen some p oposed ma ke se ings in Pales ine such as
he pe cen age o public hospi al beds (PPHB) and he
a ailabili y o p ima y heal hca e cen e s in he go e -
no a e whe e he included hospi al se es (PRC). (3) Fac-
o s conce ning he unique con ex o WB such as he
pe cen age o e ugees li ing in he go e no a e (REFP)
whe e he included public hospi al se es.
As o he i s se , six ac o s a e included. (1) The bed
occupancy a e (BOR) is ela ed o e u n o scale wi hin
hospi al ope a ions and capaci y u iliza ion, he highe he
BOR, he highe cons an e u n o scale and scale e i-
ciency [40]. F om economic poin iew, highe occupancy
a e has a lowe cos pe case [43]. (2) The a io o ou -
pa ien isi s o inpa ien days (OPIPR) shows o wha ex-
en hospi al manage s make a be e combina ion o he
wo se ices ha could make be e use o a ailable e-
sou ces. (3) The a e age leng h o s ay (ALOS) is he a e -
age days spen in a hospi al om he ime o admission o
he ime o discha ge. I ep esen s he in ensi y and e i-
ciency by which indi idual pa ien s a e ea ed [22]. (4)
The a io o adminis a i e employees o heal h employees
including doc o s (ADHR) may a ec he way o doing
clinical and nonclinical p ocesses du ing hospi aliza ion,
acco dingly may in luence e iciency [39]. (5) The size
(SIZE) o he hospi al and he applied p ocesses o pa ien
ea men may di e as o hei size and a ec he le el o
esou ce u iliza ion; a la ge hospi al may su e disecon-
omies o scale [44]. (6) Al hough all he in es iga ed hos-
pi als a e public and don’ compe e, i was el ha he
ma ke cha ac e is ics o each egion may impac e i-
ciency [33]. The p oposed ac o s in luence pa ien s’
choices and may in luence hospi al e iciency. Due o di -
e ences in demog aphic and socioeconomic ac o s, we
conside ed he loca ion o he hospi al (LOC) as a dummy
a iable o indica e whe he he hospi al is No h o
Je usalem o Sou h o Je usalem whe e di e en social
li es yles apply.
As o he second se o en i onmen al ac o s, wo
concen a ion indica o s as a p oxy o p o ide dis ibu-
ion we e included: (7) The a ailable numbe o p ima y
heal h cen e s pe 10,000 ci izens in each go e no a e
(PRC). (8) The pe cen age o public hospi al beds
(PPHB) o he o e all p o ide s’beds in a ce ain go e -
no a e [45].
As o he hi d se o en i onmen al ac o s, add-
i ional wo ac o s a e included. They apply o he
unique con ex o Pales ine. (9) Since he Pales inians’
loss o hei land and homes in 1948, ens o housands
Fig. 1 Concep ual p oduc ion s uc u e o hospi als
Table 3 Po en ial con ex ual ac o s
Va iable De ini ion Measu emen Mean
a
SD
BOR Bed occupancy a e The p opo ion o occupied beds in a
yea = Inpa ien days / (numbe o beds
a
365).
83.1% 1.65%
OPIPR Ou pa ien –inpa ien a io To al ou pa ien isi s di ided by o al inpa ien days. 1.23 0.06
ALOS The a e age leng h o s ay To al inpa ien days di ided by he numbe o admissions. 2.18 0.05
ADHR The a io o adminis a i e
o heal h employees.
The numbe o adminis a i e FTEs di ided by he o al
heal h FTEs including doc o s in each obse ed hospi al.
0.36 0.01
SIZE Hospi al size (dummy) (1) Fo la ge hospi als > 130 beds, (0) o he wise. 0.36 0.06
LOC Hospi al loca ion (dummy) (1) o No h Go e no a e and (0) o Sou h Go e no a e. 0.6 0.06
REFP The p opo ion o e ugees
li ing in he go e no a e.
The pe cen age o e ugees li ing in camps o all he
go e na e popula ion whe e he obse ed hospi al ope a es.
8% 0.8%
HPFP Numbe o hospi al beds
pe 10,000 inhabi an s
The numbe o all he a ailable hospi al beds pe 10,000
in he go e no a e whe e he public hospi al ope a es.
12 0.711
PRC The a ailable p ima y ca e
cen e s pe 10,000 inhabi an s
The numbe o p ima y cen e s pe 10,000 inhabi an s
in he go e no a e whe e he public hospi al ope a es.
2.7 0.1
PPHB The pe cen age o public
hospi al beds.
The pe cen age o he a ailable public hospi al beds in a
go e no a e o he o al a ailable numbe o beds.
59.3% 2.8%
a
Mean and SD Values used six-yea da a o he p edic o s om 2010 o 2015
Da a Sou ce: Pales inian annual heal h epo s 2010–2015
Sul an and C ispim BMC Heal h Se ices Resea ch (2018) 18:381 Page 5 o 17

o Pales inians we e displaced o li e in e ugee camps in
Wes Bank ci ies; his ac o was hough o in luence e -
iciency; he way how e ugees a e li ing and wo king
may in luence he e iciency o he wo king hospi al in
ha go e no a e. The e o e, he pe cen age o e ugees
li ing in e e y go e no a e (REFP) was conside ed. (10)
The Pales inian heal hca e sys em comp ises agmen ed
heal hca e p o ide s e ol ed h ough di e en egimes,
he numbe o a ailable hospi al beds pe 10,000 inhabi-
an s in a gi en adminis a i e a ea was conside ed I
ep esen s he supply side o hospi al se ices in a go e -
no a e (HBFP). Table 3displays he en p oposed en i -
onmen al ac o s.
Two-s age da a en elopmen analysis (2-DEA)
The p oblem o measu ing p oduc i e e iciency was bes
desc ibed, 60 yea s ago, by Fa ell [46]. To sol e he p ob-
lem, Fa ell in oduced an ac i i y analysis app oach ha
combines he measu emen o mul iple inpu s in o a single
measu e o e iciency which he ega ded as “ echnical e i-
ciency.”Technical ine iciency is he amoun o was e ha
can be elimina ed wi hou wo sening any inpu o ou pu .
Building on Fa ell’sideas,Cha nese al.[23] in oduced a
powe ul nonpa ame ic me hodology o assess he ela i e
e iciencies o mul i-inpu and mul i-ou pu p oduc ion
uni s such as hospi als which had been i led Da a En elop-
men Analysis [47]. These p oduc ion uni s a e deno ed as
decision-making uni s (DMUs) in he DEA li e a u e.
The i s published DEA wo k in heal hca e con ex was
in 1983 and in es iga ed nu sing se ices [48]. In 1984 he
second published s udy in es iga ed he medical and su -
gical depa men s in se en hospi als [49]. Among he em-
pi ical s udies using DEA, hospi als ecei ed he mos
esea ch a en ion [50]. The goals o hospi al se ices a e
mul iple and complex. Hospi als p oduce mul iple ou pu s
(e.g., inpa ien ca e, su ge ies, ou pa ien ca e, eme gency)
and abso b mul iple inpu s (e.g., clinical and non-clinical
s a , beds, equipmen , and supplies).
Based on a e iew o 317 published s udies on on ie
measu emen o he e iciency o he heal hca e deli e y
om1983 o 2006, Hollingswo h [51] ound ha 75% o
he wo ks applied he DEA, and o he DEA–based
me hods. Empi ical applica ions o DEA included pe -
o mance examina ions o di e en heal hca e ma ke s
anging om p ima y heal hca e le el [20,52] o home
heal hca e agencies [53] and hospi als [54]. And om
p ac ice beha io a p o ide g oup le el was also exam-
ined [55,56] o he o e all heal hca e sys em and coun-
y le el [45,57].
The wo-s age DEA is commonly used in p oduc i e
e iciency analysis o es ima e he impac o en i onmen-
al ac o s and p ac ices on pe o mance. Because he
DEA e iciency es ima es o he i s s age ep esen cen-
so ed da a, he second s age o analysis applies Tobi
eg ession [28,33]. Tobi eg ession applies he Max-
imum Likelihood Es ima o (MLE) o ind he model’s
pa ame e s [58]. The second s age gene a es addi ional
in o ma ion on manage ial pe o mance i we il e he
impac o he componen associa ed wi h he con ex ual
ac o s. Fu he , he second s age analysis in o ms pol-
icymake s who may in luence he ope a ing en i onmen
[14,15].
Many s udies used he DEA e iciency sco e in he sec-
ond s age analysis o e alua e he in luence o ope a ing
en i onmen on e iciency. Chowdhu y & Zelenyuk [42]
applied DEA and unca ed eg ession model o explo e
he de e minan s o he hospi al e iciency in On a io/
Canada. Thei indings iden i ied occupancy a e,
ou pa ien -inpa ien a io, loca ion, eaching s a us, and
case-mix index as de e minan s o e icien p ac ices. A
s udy examined he hospi als in Ghana used DEA and
Tobi eg ession, e iciency was de e mined by egion
and owne ship [59]. Finally, Samu & Ca ı[45] analyzed
he heal hca e sys ems in 29 OECD coun ies du ing
2000–2010 and applied Malmquis Index and Tobi e-
g ession p ocedu es, The au ho s, iden i ied educa ion,
income, and ma ke ac o s as de e minan s o hospi al
e iciency.
Despi e he ex ensi e body o DEA li e a u e examin-
ing he pe o mance o heal hca e sec o a all le els,
due o he sca ci y o da a, ew empi ical s udies we e
conduc ed in de eloping coun ies. Mos DEA wo ks
we e applied in he de eloped coun ies, mainly he US
and Eu ope [60]. Pa icula ly, in A abic Speaking Coun-
ies, wo p e ious s udies employed he DEA and in es-
iga ed he e iciency o hospi als in Jo dan and
Sul ana e o Oman [38,61]. Aimed a Pales ine, o da e,
no s udies ha e examined he e iciency o Pales inian
hospi als o he in luencing con ex ual ac o s. The pe -
o mance measu emen sys ems a e al eady absen
wi hin he coun y’s heal hca e o ganiza ions.
Me hods
This wo k add esses he p oduc i e e iciency o he
Pales inian public hospi als om 2010 o 2015. We
ex ac ed he ele an ope a ional da a om he pub-
lished Annual Heal h Repo s by he Pales inian Min-
is y o Heal h (PMoH). To achie e ou esea ch
objec i es, we o ganized he analysis a ound wo key
s eps: (1) Using a six-yea da a o he Pales inian pub-
lic hospi als, we employ he basic DEA-CCR and he
DEA-BCC models o analyze he o e all e iciency,
pu e echnical e iciency and scale e iciency; (2) we
eg ess he DEA-CCR sco es o 66 obse a ions o
he i s s ep on en po en ial con ex ual ac o s. We
apply Tobi eg ession o ind he ac o s whose im-
pac on e iciency is s a is ically signi ican .
Sul an and C ispim BMC Heal h Se ices Resea ch (2018) 18:381 Page 6 o 17
Sample and da a
The s udy used da a om 11 public hospi als ope a ing
in Wes Bank om 2010 o 2015 a ge ing a o al o 66
obse a ions. The sample excluded wo public hospi als
om he analysis. One psychia ic hospi al in Be hlehem
(180 beds) does no mee he homogenei y assump ion
o DEA me hod. Ano he newly es ablished hospi al in
2014 (37 beds) was also excluded because e iciency
judgmen o a new hospi al could be biased in he ea ly
s ages o manage ial expe ience. We ob ained e hical ap-
p o al om he Pales inian Minis y o Heal h (PMoH)
o ca y ou he esea ch.
The in es iga ed hospi als (1377 beds) a e owned and
adminis e ed by he Pales inian Minis y o Heal h. They
a e gene al hospi als and hei esou ces a e assigned om
he minis y based on eques s om hei manage s. Thei
pa ien s a e co ed by a go e nmen al insu ance scheme
by which pa ien s a e en i led o public hospi als. Then,
pa ien s a e ea ed wi hin he public hospi al unde wo
condi ions; he a ailabili y o he equi ed clinical se ices
and he a ailabili y o unoccupied hospi al bed, o he wise,
he pa ien is ans e ed o o he p o ide and inancially
co e ed by he applied insu ance scheme. Hospi al man-
age s a e asked o manage he gi en demand while man-
aging he hospi als’ esou ces acco dingly.
Public hospi als in WB a e geog aphically dis ibu ed
ac oss en go e no a es (see Table 2); one public hospi al
se es one go e na e. Heb on and Nablus a e wo excep-
ions whe e wo hospi als se e in each go e no a e. Da a
on ou inpu measu es and h ee ou pu measu es ha e
been ex ac ed om he Annual S a is ical Heal hca e Re-
po s published by he PMoH. Table 4illus a es he
yea -speci ic means and s anda d de ia ions o he in-
cluded inpu -ou pu measu es.
Es ima ion o p oduc i e e iciency
We employ wo miles ones DEA models, namely he
CCR [23] and he BCC [62]. The le e s in “CCR”and
“BCC”s and o he ini ials o he de elope s’las
names. These wo models ha e become s anda ds in he
li e a u e o pe o mance measu emen unde he as-
sump ions o cons an and a iable e u ns o scale e-
spec i ely [63]. Because public hospi als se e he public
demand as gi en and mus manage hei esou ces ac-
co dingly, he e o e, hey a ge inpu minimizing a he
han ou pu maximiza ion which ecommends using he
inpu -o ien ed DEA models [35,64–66]. We add ess he
po en ial inpu sa ings and cons uc s inpu -o ien ed
on ie s guided by he space o manage s’con ol.
Fi s , we applied a DEA-CCR model which assumes a
Cons an Re u ns o Scale (CRS) wi hin hospi al ope a-
ions and doesn’ accoun o he scale e ec s; hen, we
applied he DEA-BCC model which was de eloped in
1984 o sa is y scale e ec s in e iciency analysis. The
ma hema ical o mula ion CCR dual linea p og amming
model o es ima e ela i e e iciencies o 11 hospi als is
w i en as he ollowing linea p oblem:
θ
o¼Minθoð1Þ
Subjec o,
Xp¼11
p¼01λpxip ≤θxio i¼1;2;3;4
Xp¼11
p¼01λpy p ≥y o ¼1;2;3
λp≥0p¼1;2::; 11
Whe e:
θ
o
= he e iciency sco e o hospi al “0”unde
e alua ion.
Table 4 Dis ibu ion o inpu -ou pu measu es, means and s anda d de ia ions, N=11
Yea Inpu measu es Ou pu measu es
Hospi al beds
(X1)
Doc o s FTEs
(X2)
Heal h FTEs
(X3)
Adminis a i e FTEs
(X4)
Inpa ien days
(Y1)
Ou pa ien isi s
(Y2)
Eme gency ca e
(Y3)
2010 107 55 164 74 32,152 38,111 56,082
(20) (8) (28) (9) (6667) (7702) (7452)
2011 106 46 170 75 32,101 35,085 56,872
(19) (8) (29) (9) (6728) (7491) (7915)
2012 111 47 174 75 36,015 41,305 65,094
(21) (7) (29) (8) (7914) (8430) (9461)
2013 119 44 179 75 37,719 40,983 66,301
(23) (6) (27) (7) (8678) (8117) (11292)
2014 123 46 195 77 39,908 41,737 69,016
(25) (6) (32) (8) (9837) (8652) (11284)
2015 125 49 194 74 42,692 46,017 68,425
(25) (7) (32) (8) (10588) (8994) (10272)
FTEs Full-Time Employees. Heal h FTEs, medical pe sonnel o he han doc o s, such as nu ses, labo a o y echnicians, and adiology echnicians
Sul an and C ispim BMC Heal h Se ices Resea ch (2018) 18:381 Page 7 o 17
x
ip
= he quan i y o inpu “i”u ilized by he “p
h
”
hospi al.
y
p
= he quan i y o ou pu “ ”p oduced by he “p
h
”
hospi al.
λ= weigh s ob ained om he dual e sion o he lin-
ea p og amming.
The adial dis ance o on ie p o ides a echnical e -
iciency measu e o hospi als unde assessmen . The
DEA-BCC inpu -o ien ed model equi es an addi ional
se o con exi y cons ain o he dual linea p og am-
ming algo i hm (Eq. 1), he sum o lambdas o be one
and w i en as Eq. 2:
Xp¼11
p¼01λp¼1:0ð2Þ
The sum o lambdas yielded om he CCR model p o-
ides in o ma ion whe he he hospi al is ope a ing
unde inc easing o dec easing e u ns o scale [67,68].
While he CCR e icien hospi als a e ope a ing a he
mos p oduc i e scale size and he sum o lambdas is
one, he ine icien hospi als a e ope a ing unde De-
c easing Re u ns o Scale (DRS) when ∑λ> 1 and may
bene i om economies o scale. O he ine icien hospi-
als a e ope a ing unde Inc easing Re u ns o Scale
(IRS) when ∑λ< 1 and may su e diseconomies o scale
ha may explain a s a e o weak con ol among la ge
hospi als.
Since he BCC model always en elops he da a mo e
closely han he CCR model (inpu -o ien ed on ie s).
Ine icien hospi als measu e he sho e dis ance o he
BCC on ie han he CCR on ie [69]. The analysis o
he wo models dis inguishes h ee ypes o e iciencies
ha help manage s o cap u e he componen s o ine i-
cien ope a ions [70,71]. They a e global echnical e i-
ciency (TE) as gi en by CCR sco e, pu e echnical
e iciency (PTE) as gi en by he BCC sco e, and scale e -
iciency (SE) e lec s he po ion o ine iciency a ib-
u ed o he gi en scale o ope a ions (Eq. 3):
CCRsco e ¼BCCsco e Scale e iciency
TE ¼PTE SE ð3Þ
Rep oducing he g aph o Banke e al., [62], Fig. 2il-
lus a es he applica ion o he CCR and BCC sco es e-
ga ding he h ee componen s o e iciency ela ed o he
p oposed p oduc ion possibili y se o he inpu -ou pu
mix (X, Y). G oup o Hospi als “H1 o H6 and Hx”we e
used o demons a ion pu pose. The ine iciency com-
ponen o hospi al Hx as gi en by he a io AB/AD is a -
ibu ed o he scale o i s ope a ions. Mo eo e , i is
dis inguished om he pu e echnical ine iciency as
gi en by he a io AC/AD.
Because i is impo an o ha e a su icien numbe o
obse a ions we employed he DEA amewo k p e-
sen ed by Bousso iane e al. [71]. The me hod allows us
o cap u e he ac ual a ia ions o each hospi al h ough
simul aneous es ima ion o e iciency o all he 66 obse -
a ions (N= 66). This me hod s eng hens he disc imin-
a o y powe o DEA as su icien numbe o DMUs a e
analyzed [34].
Fig. 2 Illus a ion o SE de i ed om he CCR sco es and he BCC sco es. Rep oduced om Banke e al. [62]
Sul an and C ispim BMC Heal h Se ices Resea ch (2018) 18:381 Page 8 o 17
DEA is a ela i e measu emen me hod, a change in
he e iciency sco e in he ollowing yea o he es ed
hospi al does no necessa ily mean a change in i s pe -
o mance only; changes in he pe o mance o he o he s
may in luence he ela i e posi ion o ha hospi al. I we
ca y ou an independen analysis o each yea , we can-
no ce ainly a ibu e he changes in he e iciency sco e
o a ocal hospi al o ac ual pe o mance change o ha
hospi al. Bu , simul aneous inclusion o 66 obse a ions
in he model allows o add essing he a ia ions o a
hospi al ac oss wo successi e yea s wi h ce ain y [72].
E alua ing he impac o con ex ual ac o s on e iciency
Con ex ual ac o s which could in luence he e iciency o
a hospi al (e.g., go e nmen egula ions, geopoli ical con-
ex , ..) a e no unde he con ol o he manage and can
be accommoda ed in a DEA analysis [73]. The impac o
en i onmen on p oduc ion was i s conside ed by
Cha nes e al. [74]. The au ho s disen angled p og am e i-
ciency om managemen e iciency by e e ence o empi -
ical obse a ions ob ained om school p og ams.
F ied e al. [75] e iewed p e ious app oaches o in-
co po a ing he ex e nal ope a ing en i onmen in o a
non-pa ame ic measu e o echnical e iciency. Th ee
ca ego ies classi ied by he applied me hod in he DEA
li e a u e a e:
(1) The on ie sepa a ion app oach: can be
implemen ed only o ca ego ical ac o s and
equi es a p io i selec ion o he mos impo an
con ex ual ac o [74].
(2) The all-in-one app oach: Single-s age DEA es ima ion
o he e ec s o con ex ual ac o s had been de eloped
by Banke & Mo ey [76]. The p ocedu e includes he
ex e nal ope a ing en i onmen a iables di ec ly in
he linea p og amming p oblem along wi h he
adi ional inpu s and ou pu s. Howe e , his
app oach equi es ha he ex e nal a iable is
classi ied as an inpu o an ou pu in ad ance.
Camanho e al. [77] p opose a model ha
dis inguishes be ween he in luence o in e nal
nondisc e iona y ac o s and ex e nal
nondisc e iona y ac o s o es ima e ine iciency.
(3) The wo-s age app oach: The ypical wo-s age
app oach ollows a i s s age DEA es ima ion o
e iciency based on inpu s and ou pu s, hen a
second s age eg ession analysis seeking o explain
a ia ion in i s s age e iciency sco es conce ning
en i onmen al ac o s. Some s udies apply O dina y
Leas Squa es (OLS) eg ession o es ima e he
signi ican in luence o con ex ual ac o s in he
second s age; o he s use a Tobi eg ession model
[78]. Ray [31,79] was he i s o apply he wo-
s age DEA model whe e he es ima ed e iciency
sco es in he i s s age a e eg essed on con ex ual
a iables in he second s age.
Despi e a la ge numbe o use ul applica ions o he
wo-s age DEA me hod [29,45,80], i has been c i i-
cized and di e en examina ions o he s a is ical
consis ency o he me hod p o ided con as conclusions
ha call o u he es ing [15]. Banke & Na a ajan [14]
show by simula ion ha he wo-DEA es ima o o he
con ex ual a iables is s a is ically consis en when OLS
o Maximum Likelihood Es ima o (MLE) is applied in
he second s age. This me hod equi es he con ex ual
ac o s o be independen o he inpu a iables, bu he
con ex ual ac o s may be co ela ed wi h each o he .
Ho [28] concluded ha Tobi eg ession is su icien o
ep esen he second s age DEA models when compa ed
wi h al e na i e me hods o wi h he OLS. McDonald
[27] came o a simila conclusion as Ho , bu he ad o-
ca ed no using Tobi eg ession.
Kieschnick & McCullough [81] ecommended using
pa ame ic eg ession a he han using quasi-MLE un-
less he sample size is la ge enough o jus i y he a gu-
men unde lying he quasi-MLE. Sima & Wilson [82]
had sha ply c i icized he wo-DEA me hod o lack o a
cohe en da a gene a ing p ocess (DGP) and o he bias
and se ial co ela ion o he DEA e iciency es ima es.
They a gue ha he con en ional me hods o s a is ical
in e ence a e in alid in he second s age eg ession.
Then, he au ho s p opose he use o a boo s ap
me hod o co ec o he small sample bias and se ial
co ela ion o he DEA e iciency es ima es. La e , Da -
aio, e al. [83] es ed he assump ions equi ed o
wo-s age es ima ion and ejec ed hem in he
non-pa ame ic se ing.
We ollow Banke & Na a ajan [14]and eg ess he
DEA-CCR es ima es o 66 obse a ions du ing 2010–
2015 on en po en ial con ex ual ac o s. We un
Tobi eg ession models o iden i y which en i on-
men al ac o s ha e a signi ican in luence on he p o-
duc i e e iciency o Pales inian hospi als. The
eg ession model has a censo ed s uc u e because
he dependen a iable yielded om DEA-CCR model
is limi ed be ween ze o and one, while he independ-
en a iables ha co espond o one can be obse ed.
Then, Tobi eg ession which akes he censo ed
s uc u e in o accoun is sugges ed. The model sup-
poses ha he e is a la en dependen a iable Y
p*
,
his unobse ed a iable linea ly depends on he inde-
penden a iables X
p
ia a se o pa ame e s βs. The e
is a no mally dis ibu ed e o e m ε
p
o cap u e an-
dom in luences on he ela ion. The obse ed alue o
he dependen a iable Yp (Eq. 4) is de ined o equal
he “la en a iable”whene e he la en a iable is
abo e ze o, and o equal “ze o”o he wise, whe e:
Sul an and C ispim BMC Heal h Se ices Resea ch (2018) 18:381 Page 9 o 17
24. Em ouznejad A, Yang GL. A su ey and analysis o he i s 40 yea s o
schola ly li e a u e in DEA: 1978–2016. Socio Econ Plan Sci. 2018;61:4–8.
25. Chen Y, Wang J, Zhu J, She man HD, Chou S-Y. How he g ea ecession a ec s
pe o mance: a case o Pennsyl ania hospi als using DEA. Ann. Ope . Res. 2017:1–23.
26. Linh Pham T. E iciency and p oduc i i y o hospi als in Vie nam. J Heal h
O gan Manag. 2011;25:195–213.
27. McDonald J. Using leas squa es and obi in second s age DEA e iciency
analyses. Eu . J. Ope . Res. 2009;197:792–8.
28. Ho A. Second s age DEA: compa ison o app oaches o modelling he
DEA sco e. Eu . J. Ope . Res. 2007;181:425–35.
29. Chilinge ian JA. E alua ing physician e iciency in hospi als: a mul i a ia e
analysis o bes p ac ices. Eu J Ope Res. 1995;80:548–74.
30. PCBS PCB o S. Pales inian Heal h Accoun s 2015. 2017 [ci ed 2017 May 19].
A ailable om: h p://www.pcbs.go .ps/Downloads/book2253.pd
31. Ray SC. Resou ce-use e iciency in public schools: a s udy o Connec icu
da a. Manage. Sci. 1991;37:1620–8.
32. Ho ma che MM, Pa e son I, Riedel M. Measu ing hospi al e iciency in
Aus ia–a DEA app oach. Heal h Ca e Manag. Sci. 2002;5:7–14.
33. Na cıHÖ, Ozcan YA, Şahin İ, Ta can M, Na cıM. An examina ion o
compe i ion and e iciency o hospi al indus y in Tu key. Heal h Ca e
Manag Sci. 2015;18:407–18.
34. Cook WD, Tone K, Zhu J. Da a en elopmen analysis: p io o choosing a
model. Omega. 2014;44:1–4.
35. Özcan Yaşa A. Heal h Ca e Benchma king and Pe o mance E alua ion: An
Assessmen Using Da a En elopmen Analysis. 2007.
36. Chowdhu y H, Zelenyuk V, Lapo e A, Wodchis WP. Analysis o p oduc i i y,
e iciency and echnological changes in hospi al se ices in On a io: how
does case-mix ma e ? In J P od Econ. 2014;150:74–82.
37. Kawaguchi H, Tone K, Tsu sui M. Es ima ion o he e iciency o Japanese
hospi als using a dynamic and ne wo k da a en elopmen analysis model.
Heal h Ca e Manag Sci. 2014;17:101–12.
38. Al-Shamma i M. A mul i-c i e ia da a en elopmen analysis model o
measu ing he p oduc i e e iciency o hospi als. In J Ope P od Manag.
1999;19:879–91.
39. Kalho R, Amini S, Sokhan a M, Lo i F, Sha i i M, Kakemam E. Fac o s
a ec ing he echnical e iciency o gene al hospi als in I an: da a
en elopmen analysis. J Egyp Public Heal h Assoc. 2016;91:20–5.
40. Mujasi PN, Asbu EZ, Puig-Junoy J. How e icien a e e e al hospi als in
Uganda? A da a en elopmen analysis and obi eg ession app oach. BMC
Heal h Se . Res. 2016;16:230.
41. Ozgen H, Ozcan YA. Longi udinal analysis o e iciency in mul iple ou pu
dialysis ma ke s. Heal h Ca e Manag. Sci. 2004;7:253–61.
42. Chowdhu y H, Zelenyuk V. Pe o mance o hospi al se ices in On a io: DEA
wi h unca ed eg ession app oach. Omega. 2016;63:111–22.
43. Fe ie GD, Valdmanis V. Ru al hospi al pe o mance and i s co ela es. J
P oduc Anal. 1996;7:63–80.
44. Gok MS, Sezen B. Analyzing he ambiguous ela ionship be ween
e iciency, quali y and pa ien sa is ac ion in heal hca e se ices: he
case o public hospi als in Tu key. Heal h Policy (New. Yo k). 2013;111:
290–300. A ailable om: h p://www.sciencedi ec .com/science/a icle/
pii/S0168851013001358
45. Samu PK, Ca ıR. Analysis o he e iciency de e minan s o heal h sys ems
in OECD coun ies by DEA and panel obi . Soc Indic Res. 2016;129:113–32.
46. Fa ell MJ. The measu emen o p oduc i e e iciency. J R S a Soc Se A.
1957:253–90.
47. Cook WD, Sei o d LM. Da a en elopmen analysis (DEA)– hi y yea s on. Eu
J Ope Res. 2009;192:1–17.
48. Nunamake TR. Measu ing ou ine nu sing se ice e iciency: a compa ison
o cos pe pa ien day and da a en elopmen analysis models. Heal h Se .
Res. Heal h Resea ch & Educa ional T us . 1983;18:183.
49. She man HD. Hospi al e iciency measu emen and e alua ion: empi ical
es o a new echnique. Med Ca e. 1984;22:922–38.
50. Chilinge ian JA, She man HD. Heal h-ca e applica ions: om hospi als o
physicians, om p oduc i e e iciency o quali y on ie s. Handb da a En el
Anal. 2011:445–93.
51. Hollingswo h B. The measu emen o e iciency and p oduc i i y o heal h
ca e deli e y. Heal h Econ Wiley Online Lib a y. 2008;17:1107–28.
52. Oikonomou N, Toun as Y, Ma iolis A, Soulio is K, A hanasakis K, Ky iopoulos
J. Measu ing he e iciency o he G eek u al p ima y heal h ca e using a
es ic ed DEA model; he case o sou he n and wes e n G eece. Heal h
Ca e Manag. Sci. 2016;19:313–25.
53. Valdmanis VG, Rosko MD, Leleu H, Mukamel DB. Assessing o e all, echnical,
and scale e iciency among home heal h ca e agencies. Heal h Ca e Manag.
Sci. 2017;20:265–75.
54. Ma anga D, Bono F, Casuccio A, Fi enze AL, Ma sala L, Giaimo R, Sapienza
FR, Vi ale F. E alua ing he e ec o o ganiza ion and con ex on echnical
e iciency: a second-s age DEA analysis o I alian hospi als. Epidemiol Bios a
Public Heal. 2014:11(1).
55. Chilinge ian JA, Gla in M, Bhalo a S. Using DEA o p o ile ca diac su geon
e iciency. In: D a o Technical Repo o AHRQ; 2002.
56. Fiallos J, Pa ick J, Michalowski W, Fa ion K. Using da a en elopmen analysis
o assessing he pe o mance o pedia ic eme gency depa men
physicians. Heal h Ca e Manag. Sci. 2015:1–12.
57. Ozcan YA, Khushalani J. Assessing e iciency o public heal h and medical
ca e p o ision in OECD coun ies a e a decade o e o m. Cen Eu J Ope
Res. 2017;25:325–43.
58. Schnedle W. Likelihood es ima ion o censo ed andom ec o s. Econom
Re Taylo & F ancis. 2005;24:195–217.
59. Jehu-Appiah C, Sekidde S, Adjuik M, Akazili J, Almeida SD, Nyona o F, e al.
Owne ship and echnical e iciency o hospi als: e idence om Ghana using
da a en elopmen analysis. Cos E . Resou Alloc. 2014;12:9.
60. O’Neill L, Raune M, Heidenbe ge K, K aus M. A c oss-na ional compa ison
and axonomy o DEA-based hospi al e iciency s udies. Socio Econ Plan Sci.
2008;42:158–89.
61. Ramana han R. Ope a ions assessmen o hospi als in he Sul ana e o Oman.
In J Ope P od Manag Eme ald G oup Publishing Limi ed. 2005;25:39–54.
62. Banke RD, Cha nes A, Coope WW. Some models o es ima ing echnical
and scale ine iciencies in da a en elopmen analysis. Manage. Sci. 1984;30:
1078–92.
63. Despo is DK, Ko onakos G, So i os D. Composi ion e sus decomposi ion in
wo-s age ne wo k DEA: a e e se app oach. J. P oduc . Anal. 2016;45:71–87.
64. Lindlbaue I, Sch eyögg J, Win e V. Changes in echnical e iciency
a e quali y managemen ce i ica ion: a DEA app oach using
di e ence-in-di e ence es ima ion wi h gene ic ma ching in he
hospi al indus y. Eu . J. Ope . Res. 2016;250:1026–36.
65. Mogha SK, Yada SP, Singh SP. Es ima ing echnical e iciency o public
sec o hospi als o U a akhand (India). In J Ope Res Inde science
Publishe s (IEL). 2016;25:371–99.
66. Wang C, Wang X, Su Q, Du J. How can hospi als pe o m mo e e icien ly? A
case s udy in China based on da a en elopmen analysis. Se . Sys . Se .
Manag. (ICSSSM), 2015 12 h In . Con . IEEE; 2015. p. 1–4.
67. Banke RD, Th all RM. Es ima ion o e u ns o scale using da a en elopmen
analysis. Eu . J. Ope . Res. 1992;62:74–84.
68. Banke RD, Mo ey RC. E iciency analysis o exogenously ixed inpu s and
ou pu s. Ope Res. 1986;34:513–21.
69. Dyson RG, Allen R, Camanho AS, Podino ski VV, Sa ico CS, Shale EA. Pi alls
and p o ocols in DEA. Eu . J. Ope . Res. 2001;132:245–59.
70. Coope WW, Sei o d LM, Tone K. Da a En elopmen Analysis: A
Comp ehensi e Tex wi h Models, Applica ions, Re e ences and DEA-Sol e
So wa e. 2000.
71. Bousso iane A, Dyson RG, Thanassoulis E. Applied da a en elopmen
analysis. Eu . J. Ope . Res. 1991;52:1–15.
72. Sa kis J, Tallu i S. E iciency measu emen o hospi als: issues and ex ensions.
In . J. Ope . P od. Manag. 2002;22:306–13.
73. Coelli TJ, Rao DS, O’Donnell CJ, Ba ese GE. An in oduc ion o e iciency
and p oduc i i y analysis: Sp inge Science & Business Media; 2005.
74. Cha nes A, Coope WW, Rhodes E. E alua ing p og am and manage ial
e iciency: an applica ion o da a en elopmen analysis o p og am ollow
h ough. Manage Sci. 1981;27:668–97.
75. F ied HO, Schmid SS, Yaisawa ng S. Inco po a ing he ope a ing
en i onmen in o a nonpa ame ic measu e o echnical e iciency. J
P oduc Anal. 1999;12:249–67.
76. Banke RD, Mo ey RC. The use o ca ego ical a iables in da a en elopmen
analysis. Manage. Sci. 1986;32:1613–27.
77. Camanho AS, Po ela MC, Vaz CB. E iciency analysis accoun ing o
in e nal and ex e nal non-disc e iona y ac o s. Compu Ope Res. 2009;
36:1591–601.
78. Adle N,F iedmanL,Sinuany-S e n Z. Re iew o anking me hods in
he da a en elopmen analysis con ex . Eu . J. Ope . Res. 2002;140:
249–65.
79. Ray SC. Da a en elopmen analysis, nondisc e iona y inpu s and e iciency:
an al e na i e in e p e a ion. Socioecon Plann Sci. 1988;22:167–76.
Sul an and C ispim BMC Heal h Se ices Resea ch (2018) 18:381 Page 16 o 17

80. B a o-U e a BE, Solís D, López VHM, Ma ipani JF, Thiam A, Ri as T. Technical
e iciency in a ming: a me a- eg ession analysis. J. P oduc . Anal. 2007;27:
57–72.
81. Kieschnick R, McCullough BD. Reg ession analysis o a ia es obse ed on (0,
1): pe cen ages, p opo ions and ac ions, ol. 3. Thousand Oaks: S a
Modelling Sage Publica ions Sage CA; 2003. p. 193–213.
82. Sima L, Wilson PW. Es ima ion and in e ence in wo-s age, semi-pa ame ic
models o p oduc ion p ocesses. J Econom. 2007;136:31–64.
83. Da aio C, Sima L, Wilson PW. Tes ing whe he wo-s age es ima ion is
meaning ul in non-pa ame ic models o p oduc ion. ISBA discuss. Pap.
Google Sch. 2010;
84. Amemiya T. Reg ession analysis when he dependen a iable is unca ed
no mal. Econom J Econom Soc. 1973:997–1016.
85. Ozcan YA. Pe o mance measu emen using da a en elopmen analysis
(DEA). Heal Ca e Benchma king Pe o m E al An Assess using Da a En el
Anal Sp inge ; 2008;15–41.
86. A ilgan E. The echnical e iciency o hospi al inpa ien ca e se ices: an
applica ion o Tu kish public hospi als. Bus Econ Res J. 2016;7:203.
87. Mo isey MA. Compe i ion in hospi al and heal h insu ance ma ke s: a
e iew and esea ch agenda. Heal h Se . Res. Heal h Resea ch &
Educa ional T us . 2001;36:191.
88. Nyman JA, B icke DL. P o i incen i es and echnical e iciency in he
p oduc ion o nu sing home ca e. Re Econ S a . 1989:586–94.
89. Tlo lego N, Non ignon J, Sambo LG, Asbu EZ, Ki igia JM. Assessmen o
p oduc i i y o hospi als in Bo swana: a DEA applica ion. In A ch Med.
2010;3:27.
Sul an and C ispim BMC Heal h Se ices Resea ch (2018) 18:381 Page 17 o 17