Int
ern
at
i
onal
Journ
al of
P
u
bli
c Hea
lt
h S
c
ie
nce (
I
JPHS)
Vo
l.
6
,
No.
3
,
S
eptem
ber
201
7
, pp.
203
~
21
2
IS
S
N:
22
52
-
8806,
DOI:
10
.11
591/ij
phs.
v6
i
3.847
5
203
Journ
al h
om
e
page
:
https:
//
ww
w.i
aesco
re
.c
om
/
jo
urn
als/i
nd
ex.
php/IJP
H
S
Timi
ng
and Dete
rminant
s o
f
Tuber
culosi
s
T
reatm
ent
I
nterrup
tion in N
airobi C
ounty
, K
enya
Violet
J
e
pchu
mba
1
,
Sim
on
Ka
r
anja
2
,
Ev
ans A
muk
oye
3
, L
awrence
Mutham
i
4
,
Hil
lar
y
Kipr
uto
5
1,2
Jom
o
Ken
y
at
t
a
Univer
si
t
y
of
Agric
ult
ur
e and Te
chno
log
y
(JK
UA
T)
,
Ken
y
a
3
,4
Ken
y
a
Medical
R
ese
ar
ch
Insti
tut
e
(KEMRI)
,
Ken
y
a
5
W
orld
Hea
lt
h
Organi
zation (W
HO
)
,
Sw
it
ze
r
la
n
d
Art
ic
le
In
f
o
ABSTR
A
C
T
Art
ic
le
history:
Re
cei
ved
J
un
12
, 2
017
Re
vised A
ug 11
, 2
017
Accepte
d Aug
25
, 201
7
Tube
rcu
losis
(T
B)
treatment
is
a
ke
y
p
il
l
ar
in
th
e
m
ana
gement
an
d
cont
ro
l
o
f
TB.
Servi
ce
de
liver
y
wi
thi
n
th
e
t
rea
tment
f
ac
i
li
t
i
es
play
s
an
important
rol
e
in
ensuring
treat
m
ent
adh
ere
n
ce
b
y
TB
pa
ti
en
ts.
A
prospec
ti
v
e
c
ohort
stu
d
y
invol
ving
25
he
al
th
f
ac
i
li
t
ie
s,
2
5
fac
i
li
t
y
in
-
charge
officers
and
291
pat
ie
n
ts
dia
gnosed
as
ne
w
sputum
s
m
ea
r
positi
ve
(SM
+)
bet
wee
n
Dec
e
m
ber
2014
and
Jul
y
2015
was
under
ta
ke
n
.
The
ai
m
of
the
stud
y
was
to
esti
m
at
e
the
m
edi
an
t
ime
t
o
treatment
interrupti
on
,
associa
t
ed
fa
ct
ors
a
nd
over
a
ll
pre
dictors
of
no
n
-
adhe
ren
ce
to
TB
treatment
.
A
tot
a
l
of
19
(6
.
5
%)
treatment
int
err
up
ti
ons we
re
observe
d.
The m
edi
an
ti
m
e
to defa
ul
t
was 56
[95%
CI,
36
-
105]
day
s.
Tr
eat
m
ent
in
a
non
-
publi
c
fa
ci
l
ity
[
AO
R=0.
210,
95%
CI
(0.
046
-
0.
952)
]
and
facil
ities
per
ceive
d
to
hav
e
ad
equ
at
e
num
ber
of
hea
l
th
c
are
workers
to
offe
r
Dire
ctl
y
Obs
er
ved
Therap
y
(D
OT)
[AO
R=0.
1
95,
95%
CI
(0.
068
-
0.
56)]
show
ed
a
lower
odds
of
tre
a
t
m
ent
int
err
up
tion
where
as
at
t
ai
nm
ent
of
se
conda
r
y
l
evel
e
duca
t
ion
[AO
R=5.
28,
95%
CI
(1.
18
-
23.
59)]
indi
c
at
ed
a
h
ig
her
odds
of
treatment
int
err
up
t
ion.
Non
-
c
li
ni
cal
aspe
ct
s
of
hea
l
th
c
are
serv
ic
e
deliver
y
influence
patient
a
dher
ence
to
TB
treatment
.
Hea
lt
h
se
eki
ng
beha
vior
of
gro
ups
conside
red
t
o
be
high
r
isk
for
tre
at
m
en
t
int
err
up
ti
on
should
be
inc
orpor
at
ed
in
to
the
d
esign
and
del
iv
er
y
of
T
B
tre
a
tment
.
Ke
yw
or
d:
Ri
sk
f
act
or
s
Su
r
vi
val
a
naly
sis
T
uberc
ulo
sis
Treatm
ent
i
nterrup
ti
on
Copyright
©
201
7
Instit
ut
e
o
f Ad
vanc
ed
Engi
n
ee
r
ing
and
S
cienc
e
.
Al
l
rights re
serv
ed
.
Corres
pond
in
g
Aut
h
or
:
Violet Je
pchu
m
ba
,
Jo
m
o
Ke
n
ya
tt
a Unive
rsity
o
f
Agricult
ure
a
nd Tec
hnol
og
y
,
P.O.Bo
x 620
0
-
00200,
Nair
ob
i
, K
e
nya
.
Em
a
il
: violet
.jepchum
ba@
gm
ai
l.co
m
1.
INTROD
U
CTION
Tu
ber
c
ulo
sis
(
TB)
is
on
e
of
the
infecti
ou
s
di
seases
of
publ
ic
healt
h
co
nce
rn
globall
y.
A
ccordin
g
to
WHO
Gl
obal
TB
repor
t
(
20
16)
the
re
was
an
est
i
m
at
ed
10.4
m
il
l
ion
ne
w
(inci
de
nt)
TB
cases
worl
dw
i
de
includi
ng
480,
000
ne
w
cases
of
m
ulti
dr
ug
-
resist
ant
TB
(
MDR
-
TB)
a
nd
an
add
it
io
nal
100,0
00
peopl
e
with
rifam
picin
-
resi
sta
nt
TB
(RR
-
TB)
[
1]
.
Ke
nya
is
cur
re
ntly
r
ank
e
d
am
on
g
t
he
to
p
20
c
ountries
with
high
TB,
high
TB/
H
I
V
an
d
high
M
DR
-
TB
bur
de
n,
acc
ountin
g
for
84%,
87
%
an
d
84%
of
t
he
global
bur
de
n
resp
ect
ively
.
I
n
201
5
the
Ke
nya
Divisio
n
of
Lep
ro
sy,
T
uberc
ulo
sis
a
nd
Lung
Disease
(D
LTL
D
)
noti
fied
a
total
nu
m
ber
of
81,
518
of
w
hi
ch
74,
742
we
r
e
new
cases
w
hile
6,776
we
r
e
pr
e
viously
treat
ed
cases.
N
ai
robi
County
noti
fie
d
the
h
i
gh
e
st n
um
ber
of cases
at 1
2,385 [
2].
W
it
h
co
rr
ect
m
anag
em
ent
treatm
ent
su
ccess
rates
are
high.
If
le
ft
untreat
e
d,
TB
has
a
hi
gh
m
or
ta
li
ty
rate.
P
oor
a
dh
e
ren
ce
t
o
anti
-
T
B
drugs
can
le
ad
to
em
erg
en
ce
of
dru
g
re
sist
ant
TB.
Re
sist
ance
to
si
ng
l
e
dru
gs
has
bee
n
re
po
rted
i
n
e
ver
y
country
a
nd
r
esi
sta
nce
to
al
l
of
the
m
ajo
r
anti
-
tu
bercul
osi
s
dru
gs
has
no
w
e
m
erg
ed
[
3].
Ov
e
r
t
he
ye
ars
,
de
sp
it
e
the
de
cl
ine
in
case
fin
ding
f
or
drug
se
ns
it
ive
T
B,
Ke
nya
has
seen
a
gr
a
dual
increa
se
in
DR
-
TB
c
ase
noti
ficat
io
n
f
r
om
11
2
ca
ses
in
2010
t
o
433
in
2015.
No
ta
bly,
the
re
was
a
50% inc
rease i
n 201
5,
c
om
par
ed
to 2
88 case
s in 2
014.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2252
-
8806
IJPHS
V
ol.
6
, No
.
3
,
Septem
ber
201
7
:
203
–
21
2
204
Av
ai
la
bili
ty
of
r
eso
urces
f
or
T
B
care
a
nd
pre
ven
ti
on
is o
ne
of
t
he
c
om
po
ne
nts
in
t
he
im
ple
m
entat
ion
of
WHO
E
ND
TB
strat
egy
[4]
.
Ther
e
is
ina
dequate
num
ber
of
healt
h
care
work
e
rs
(
HC
Ws)
a
nd
sk
il
ls
acro
s
s
al
l
cadr
es
of
hum
an
resou
rce
s
f
or
healt
h
i
n
Ken
ya
.
Acc
ord
ing
t
o
Ken
ya
Ser
vice
A
vaila
bili
ty
and
Re
a
din
es
s
Assessm
ent
M
app
i
ng
[
5],
the
do
ct
or
-
popula
ti
on
rati
o
is
le
ss
than
one
(<1)
per
10,
000
popula
ti
on.
The
nu
rs
e
popula
ti
on
-
rati
o
was
est
a
blished
as
3
per
10,
000
pe
op
le
a
nd
f
or
re
gister
ed
cl
inica
l
offi
cers
1
per
10,
000.
All
oth
e
r
cad
res
of
HC
W
s
acc
ounted
f
or
le
ss
th
an
one
(<
1)
pe
r
10,
000.
T
his
is
ver
y
low
c
om
par
ed
to
the
WHO
reco
m
m
end
ed
sta
nd
a
rds
of
23
doct
ors,
nurs
es
an
d
m
idwiv
es
per
10,
000
per
s
ons
in
a
pop
ulati
on.
Si
gnific
an
t
workf
or
ce
ga
ps
are
com
plica
te
d
by
une
qu
it
able
distrib
utio
n
of
HC
Ws.
T
he
Healt
h
Sect
or
-
H
um
an
Re
so
urce
Strate
gy
(
HS
:
HRS)
2014
-
20
18
[
6]
pro
j
ect
s
that
by
the
ye
ar
20
30
t
her
e
will
be
a
chall
eng
e
i
n
fill
in
g
gen
e
ral
pr
act
it
ion
e
r
an
d
nurse
s
ga
ps
in
the
country.
Ken
ya
has
a
total
of
8,4
05
healt
h
facil
it
ie
s
of
w
hich
49%
are
op
e
rated
b
y t
he
gov
e
r
nm
ent (
public).
Of the
se 5,8
40
offer
TB t
reatm
ent serv
ic
es
[
5].
The
loss
t
o
f
ollow
up
durin
g
tub
e
rcu
l
os
is
(T
B)
treat
m
ent
c
on
t
rib
utes
to
low
treat
m
ent
su
ccess
rate
s
and
possible
de
velo
pm
ent
of
dru
g
resist
ant
TB
(D
R
-
TB).
Ken
ya
has
see
n
an
increa
se
in
cases
of
DR
-
TB
case
no
ti
ficat
io
n
w
hich
is
of
grea
t
con
ce
r
n
as
is
poses
a
t
hr
eat
to
the
gain
s
m
ade
in
re
duc
ing
TB
bur
de
n
in
the
country.
S
ucce
ssfu
l
treat
m
ent
of
dru
g
sensit
ive
TB
is
of
utm
os
t
i
m
po
rtance
in
e
ns
uri
ng
that
the
c
hances
of
dev
el
op
i
ng
D
R
-
TB
are
m
inim
iz
ed.
Adher
e
nce
to
the
six
m
on
th
treatm
e
nt
reg
im
en
fo
r
al
l
new
ly
diagnose
d
patie
nts
can
no
t
be
ov
e
rem
ph
a
siz
ed
in
the
ef
forts
to
halt
further
in
crease
in
r
ep
or
te
d
cas
es
of
MDR
T
B.
In
2015,
Ken
ya
r
ecorde
d
4.3%
t
reatm
ent
interru
ptio
n
rate
am
ongs
t
ne
w
pu
l
m
on
ary
TB
patie
nts
[2
]
.
Fact
ors
that
con
t
rib
ute
to
non
-
ad
her
e
nce
shou
l
d
be
a
ddress
ed
im
m
e
diate
ly
to
incr
ease
treat
m
ent
su
ccess
rates
of
dru
g
su
sce
ptible
TB
an
d
st
op
the
s
urge
i
n
i
ncide
nc
es
DR
-
TB.
Tr
eatm
ent
interr
upti
on
is
def
i
ne
d
as
a
patie
nt
bein
g
off
-
treat
m
ent f
or tw
o
c
on
sec
ut
ive m
on
ths.
2.
METHO
D
2.1.
Opera
tio
na
l
d
efinition
s
Treatm
ent
inter
r
up
ti
on
acc
ordin
g,
to
the
WHO
a
nd
In
t
ern
at
io
nal
U
ni
on
a
gainst
T
uberc
ulo
sis
an
d
Lu
ng
Disease
(
IU
AT
LD
)
gu
i
delines
is
def
i
ned
as
TB
patie
nt
w
ho
di
d
not
sta
rt
tr
eatm
ent
or
w
ho
se
trea
t
m
ent
was
i
nterru
pted for tw
o
c
ons
ecuti
ve
m
on
ths
or
m
or
e [
7].
Sm
ear
po
sit
ive
PTB
was
def
i
ned
as
a
patie
nt
with
at
le
ast
one
s
pu
t
um
sm
ear
exam
inatio
n
p
os
it
ive
for
aci
d
fast
ba
ci
ll
i.
Alte
rn
at
ively
,
patie
nts
whose
t
wo
sa
m
ples
turn
ed
s
m
ear
ne
gative,
wer
e
gi
ve
n
an
ti
bio
ti
cs
and
on
re
peat
diag
no
sis
one
sa
m
ple
turned
posit
ive
f
or
Myco
bact
eri
um
tu
berc
ulos
is
bacil
li
,
wer
e
al
so
cl
assifi
ed
as
S
M+
[
8].
Treatm
ent
was
adm
inist
ered
accor
ding
t
o
t
he
WHO/I
UATLD
guideli
ne
s
[9
]
.
The
pati
ents
recei
ved
a
sta
nd
a
r
dized
Shor
t
-
Co
ur
se
Chem
oth
erapy
(S
CC
)
fo
ll
owin
g
the
Dire
ct
ly
Ob
ser
ve
d
T
he
rap
y
(DO
TS)
f
or
treatm
ent
of
TB
strat
egy
.
Th
e
reco
m
m
end
ed
re
gim
ent
fo
r
treatm
ent
fo
r
ne
w
SM
+
adu
lt
s
entai
le
d
two
m
on
ths
of
inte
ns
ive
phase
treatm
ent
with
f
our
dru
gs
con
sist
in
g
of
Isoniazi
d
(
H),
Ri
fa
m
picin
(R),
P
yrazi
nam
ide
(Z)
and
Et
ham
bu
tol
(E).
This
w
as
then
f
ollowe
d
by
a
con
ti
nuat
ion
ph
a
se
of
Ri
fam
picin
a
nd
Isoniazi
d
(
RH)
f
or
four m
on
ths
.
2.2.
S
tu
d
y
site
This
stu
dy
wa
s
conduct
ed
in
Nairobi
Coun
ty
wh
ic
h
re
por
ts
the
hig
hest
CNR
of
SM+
PTB
in
the
country
[2
]
.
A
total
of
25
faci
li
ti
es
of
fe
rin
g
TB
treat
m
ent
within
Nair
obi
Co
un
ty
wer
e
inclu
ded
in
the
stud
y.
The faci
li
ti
es w
ere
sel
ect
ed base
d
on t
he hig
h n
um
ber
of
TB pati
ents
rec
ei
vin
g TB
treat
m
ent. Th
e
sam
ple size
per facil
it
y wa
s all
ocated
pro
portio
nally
b
as
ed on t
he n
umber
of S
M+
T
B pati
ents treat
ed
in
20
13.
2.3.
St
ud
y
desig
n
A
pros
pecti
ve
cohor
t
stu
dy
of
patie
nts
di
agnose
d
betw
een
Decem
ber
2014
a
nd
Ju
l
y
20
15
wa
s
unde
rtake
n.
T
he
patie
nts
w
ere
inter
viewe
d
twic
e
durin
g
the
treat
m
e
nt
pe
rio
d.
Th
e
first
inte
rv
i
ew
w
a
s
adm
inist
ered
within
t
he
fir
st
three w
ee
ks
of
bein
g
diag
nos
ed
as
SM+
.
T
he
seco
nd
i
nter
vi
ew
was
a
dm
inist
ered
after
12
wee
ks
of
treat
m
ent.
T
he
facil
it
y
in
-
charges
wer
e
i
nter
viewe
d
on
ce
within
the
s
tud
y
pe
rio
d
.
A
fter
six
m
on
ths of treat
m
ent, TB r
e
gis
te
rs
we
re
rev
ie
wed to c
ollec
t
data on t
reatm
ent outcom
es.
2.4.
St
ud
y
popula
t
ion
The
st
ud
y
pa
rtic
ipants
c
on
sti
t
uted
ne
w
SM+
PTB
patie
nts
in
Nai
robi
Co
un
ty
.
O
nl
y
pat
ie
nts
ab
ove
15
ye
a
rs
wer
e
include
d
i
n
the
stud
y
for
ease
of
sputum
colle
ct
ion
f
or
c
on
firm
ation
as
S
M+
sta
tus.
Pat
ie
nts
who
ha
d
bee
n
on
treat
m
ent
fo
r
m
or
e
than
t
hr
ee
week
s
we
re
not
include
d
in
the
stud
y.
An
in
f
or
m
ed
con
s
ent
was
sou
gh
t
f
r
om
each
par
ti
ci
pan
t
befor
e
da
ta
colle
ct
ion
.
Faci
li
ty
-
in
-
cha
rg
es
from
all
par
ti
ci
patin
g
f
aci
li
ti
e
s
wer
e
inclu
de
d
in
the
stu
dy.
T
he
hypothesis
f
or
the
sam
ple
s
iz
e
cal
culat
ion
was
base
d
on
a
n
e
xp
ect
e
d
a
dv
erse
treatm
ent
ou
tc
om
e
of
12%
i
n
new
sm
ear
po
sit
ive
TB
cas
es
an
d
a
n
a
b
sol
ute
preci
sion
of
0.0
3.
Furthe
rm
or
e
Evaluation Warning : The document was created with Spire.PDF for Python.
IJPHS
IS
S
N:
22
52
-
8806
Timin
g and De
te
rminan
ts
of T
ub
e
rcul
os
is Tr
eatme
nt Interr
up
ti
on i
n
N
airo
bi
....
(
Viole
t
Jepchu
mba)
205
the
sa
m
ple
size
was
increase
d
by
20%
for
exp
ect
e
d
losse
s
du
e
to
patie
nt
s
diagnose
d
as
s
m
ear
-
posit
ive
w
ho
did
not
ret
urn
for
treat
m
ent.
In
t
otal,
291(9
2
%
respo
ns
e
r
at
e)
ne
w
SM+
TB
patie
nts
a
nd
25
(
100%
re
sp
onse
rate)
facil
it
y i
n
cha
rg
es
w
e
re i
nter
viewe
d.
2.5.
Data
c
ollec
tio
n
Tw
o
data
sou
r
ces
wer
e
us
e
d
in
the
stu
dy.
S
tructu
red
Q
ues
ti
on
nai
res
we
r
e
adm
inist
ered
to
the
stu
dy
par
ti
ci
pa
nts
within
th
ree
wee
ks
of
s
ta
rtin
g
t
reatm
ent
and
once
durin
g
t
he
con
ti
nuat
io
n
phase
t
o
c
ollec
t
so
ci
o
-
dem
og
ra
ph
i
c
a
nd
patie
nt
cha
racteri
sti
cs.
F
aci
li
t
y
in
-
charg
e
wer
e
inter
viewed
to
c
ollec
t
data
on
insti
tuti
on
al
char
act
e
risti
cs.
Treatm
ent
ou
t
com
e(s)
we
re
extracte
d
f
ro
m
the
TB
re
giste
r
at
the
en
d
of
the
treat
m
ent
per
io
d.
The
data
c
ollec
ti
on
to
ols
were
pr
et
est
ed
to
ens
ur
e
that
the
qu
est
io
ns
were
con
sist
e
nt.
D
at
a
ver
ific
at
ion
wa
s
carried
out
to
ens
ur
e
data
va
li
dity
.
On
rec
e
ipt
of
quest
io
nnai
res,
i
nter
nal
and
exter
nal
consi
ste
ncy
ch
ecks
w
ere
ca
rr
ie
d
out.
A
d
ou
ble
entry
proce
dure
was
ad
opte
d
i
n
data
e
ntry
for
al
l
the
data.
The
tw
o
im
ag
e
data
file
s
wer
e
com
par
e
d
f
or
c
onsist
ency.
Discre
pan
t
val
ues
we
re
chec
ked
a
ga
inst
the
or
i
gin
a
l
data
file
and
value
s
and
inc
onsist
encies
wer
e
c
orrecte
d.
T
his
le
d
to
the
eve
ntu
al
creati
on
of
the
final
data
base
f
or
safe
ke
epin
g
and analy
sis.
2.6.
Data
anal
ys
is
The
c
om
pu
te
d
descr
i
ptive
sta
t
ist
ic
s
wer
e
us
e
d
to
descr
i
be
da
ta
char
act
e
risti
cs
in
te
rm
s
of
m
eans
an
d
pro
portions.
I
n
biv
a
riat
e
an
al
ysi
s,
Chi
-
s
quare
was
us
ed
to
te
st
for
a
s
so
ci
at
ion
bet
w
een
t
he
in
de
pe
nd
e
nt
var
ia
b
le
s
an
d
t
reatm
ent
interru
ptio
n.
Ka
plan
-
Me
ie
r
fail
ure
est
i
m
at
es
and
log
ra
nk
te
st
w
ere
us
ed
to
est
i
m
at
e
aver
a
ge
fail
ure
tim
e,
cor
res
pondin
g
le
vel
of
risk
a
nd
diff
e
re
nce
in
patie
nt
ave
ra
ge
tim
e
.
Trea
t
m
ent
interr
up
ti
on
for
reas
ons
ot
her
than
fact
or
un
der
a
naly
sis
w
as
cens
or
e
d
in
the
Ka
plan
-
Me
ie
r
su
r
vi
val
an
al
ysi
s.
Cum
ulati
ve
ris
k
of
treat
m
ent
interr
up
ti
on
durin
g
the
entire
treatm
ent
per
io
d
was
determ
i
ned.
Co
x
regre
ssion
hazar
d
a
naly
sis
was
use
d
t
o
determ
ine
predict
or
s
of
tre
atm
ent
interrupti
on.
Stat
ist
ic
al
sign
ific
anc
e
was
determ
ined
by
consi
der
i
ng
no
m
inal
p
-
value
of
le
ss
tha
n
5%
(
p
<
0.0
5)
wi
th
a
95%
c
onfiden
ce
le
vel.
D
at
a
wa
s
analy
zed
us
in
g St
at
a
Corp.
20
13.
Stat
a Stat
is
ti
cal
So
ftwa
re:
Re
le
ase 13. C
ol
le
ge
S
ta
ti
on, T
X: Stat
aC
orp L
P.
3.
RESU
LT
S
3
.1
.
S
oc
i
o
d
emo
gra
ph
i
c
c
ha
ra
cte
r
i
s
t
i
cs
O
f
t
he
2
91
r
es
po
nd
en
t
s
,
21
5
(7
4%
)
we
r
e
m
a
l
e
.
The
TB
patie
nts
ha
d
an
a
ve
rag
e
age
of
32
.
3
(22.3
-
42.
3)
ye
ars.
Ma
jority
of
the
patie
nts 147
(
51.
7%
)
ha
d
at
ta
ined
a
se
conda
ry
le
vel
of
e
du
cat
io
n,
99
(
35
%
)
pr
im
ary
le
vel
and
38
(
13.
3
%)
post
-
sec
ondar
y
ed
ucati
on
.
A
la
rg
e
pro
portio
n
of
the
stud
y
par
ti
ci
pa
nts
21
2
(74%
)
had
a
source
of
inc
om
e
of
w
hich
11
7
(55
%)
we
re
se
lf
-
em
plo
ye
d
w
hile
95
(
45%)
w
ere
em
plo
ye
d
by
a
third
pa
rty
.
Pe
r
so
ns
em
plo
ye
d
by
t
hir
d
par
ti
e
s
we
re
furthe
r
segr
e
gated
a
nd
cl
us
te
re
d
i
nto
gro
up
s
ba
sed
on
t
he
fr
e
qu
e
ncy
of
pa
ym
ent,
64
(
64
%
)
recei
ved
paym
ent
on
a
m
on
thly
basis,
16
(16%
)
on
a
wee
kly
basi
s
an
d
19(
19
%
)
wer
e
paid
on
a
da
il
y
basis.
The
patie
nt
s
ocio
-
dem
og
ra
ph
ic
c
har
act
er
ist
ic
s
wer
e
c
r
os
s
ta
bula
te
d
against
TB
trea
t
m
ent interr
up
t
ion
show
n i
n
T
able 1.
Table
1
.
C
ro
s
s
Tab
ulati
o
n o
f
Pati
ent
So
ci
o
-
dem
o
gr
a
phic
Ch
aracte
risti
cs Agai
ns
t
TB
Treatm
ent I
nter
r
up
ti
on
So
cio
-
d
e
m
o
g
r
ap
h
ic charact
eristics
TB T
reat
m
en
t Out
co
m
e
No
.
Cen
so
red
Tr
eat
m
en
t
Interrup
tio
n
Gen
d
er
Male
1
9
9
(92
.9
9
%)
1
5
(7.0
1
%)
Fe
m
ale
7
1
(95
.95
%)
3
(4.0
5
%)
Ag
e
15
-
24
4
7
(88
.68
%)
6
(11
.32
%)
25
-
34
1
2
7
(93
.3
8
%)
9
(6.6
2
%)
35
-
44
5
9
(93
.65
%)
4
(6.3
5
%)
4
5
plu
s
3
5
(10
0
%)
0
(0.0
0
%)
Level o
f
edu
catio
n
Pri
m
a
r
y
9
5
(96
.94
%)
3
(3.0
6
%)
Seco
n
d
ary
1
3
2
(89
.8%
)
1
5
(10
.2%
)
Po
st
-
Seco
n
d
ary
3
8
(10
0
%)
0
(0.0
0
%)
Hav
e Sou
rce
o
f
inco
m
e
Yes
1
9
6
(92
.8
9
%)
1
5
(7.1
1
%)
No
7
1
(94
.67
%)
4
(5.3
3
%)
Ty
p
e of
so
u
rce
o
f
inco
m
e
Self
e
m
p
lo
y
ed
1
1
0
(94
.0
2
%)
7
(5.9
8
%)
E
m
p
lo
y
ed
8
6
(91
.46
%
8
(8.5
1
%)
Frequ
en
cy
o
f
W
ag
e pay
m
en
t
Daily
1
6
(84
.21
%)
3
(15
.79
%)
W
eekl
y
1
2
(
8
0
.00
%)
3
(20
.00
%)
Mon
th
ly
6
2
(95
.38
%)
3
(4.6
2
%)
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2252
-
8806
IJPHS
V
ol.
6
, No
.
3
,
Septem
ber
201
7
:
203
–
21
2
206
3.2.
Pat
ie
n
t
c
ha
r
ac
teristics
At
rec
ru
it
m
ent,
136
(
46.7%)
of
the
patie
nts
r
eported
a
histo
ry
of
sm
ok
in
g
.
The
sm
ok
ers
wer
e
f
ur
t
her
cl
us
te
red,
ba
se
d
on
the
la
st
ti
m
e
they
s
m
ok
ed.
T
he
fi
nd
i
ng
s
showe
d
that
m
ajo
rity
of
the
patie
nts,
75
(
58.
7%)
had
sm
ok
ed
within
8
wee
ks
or
le
ss
to
the
day
of
the
intervi
ew.
D
ur
i
ng
the
con
ti
nuat
ion
phase,
m
ajo
rity
of
th
e
respo
nd
e
nt
s
65
(88.9%)
,
we
re
no
t
sm
ok
in
g.
History
of
al
co
ho
l
c
on
s
um
pti
on
was
re
porte
d
by
16
8
(
58.3%)
of
the
stud
y
pa
rtic
ipants
on
rec
r
uit
m
ent
into
the
stud
y.
Ma
jor
it
y
had
con
s
um
ed
al
coh
ol
w
it
hin
8
wee
ks
and
le
ss
to
the
st
udy
po
int.
D
ur
i
ng
the
co
ntinu
at
io
n
phase,
m
os
t
pati
ents
85
(89%
)
wer
e
not
c
on
s
um
ing
al
cohol.
Most
of
t
he
par
ti
ci
pa
nts
25
3(
94%
)
are
s
pen
t
one
hour
or
le
ss
tra
ve
ll
ing
to
a
cces
s
a
he
al
th
facil
it
y.
The
m
ean
travel
tim
e
to
the
fa
ci
li
t
y
repor
te
d
by
the
pa
rtic
ipants
was
1.0
7
±
0.3
hrs.
F
ro
m
our
fi
nd
i
ng
s
,
87
(
31%)
of
the
par
ti
ci
pa
nts suff
e
red
a
dv
e
rse
reacti
on
to
TB
treatm
ent w
it
hin
the f
irst t
hr
ee
w
eeks of
treat
m
ent.
Sk
in irr
it
at
ion
was
the
m
os
t
com
m
on
33(
37.
9
%
)
,
si
de
e
ff
e
ct
repor
te
d,
D
uri
ng
the
co
ntin
uation
phase
,
t
he
r
ep
or
te
d
a
dverse
reacti
on
f
reque
ncy
am
on
g
the
stu
dy
pa
rtic
ip
ants
droppe
d
t
o
49
(
19%)
of
wh
ic
h
s
kin
irri
ta
ti
on
,
17(
43%
)
was
the
m
os
t
com
m
on
.
W
it
hi
n
t
he
first
t
hr
ee
weeks
of
treat
m
ent,
reli
ef
on
TB
sym
pto
m
s
was
re
porte
d
by
157
(72%
)
of
the
st
ud
y pa
rtic
ipant
s,
Mo
st
of
t
he
m
75
(51%
)
,
no
te
d
a reducti
o
n
in
c
oughin
g.
T
he
st
ud
y pa
rtic
ipants
repor
te
d
prese
nce
of
oth
er
c
hro
nic
infecti
ons
51
(18%
)
of
wh
ic
h
H
IV
/
A
ID
S
was
t
he
m
os
t
co
m
m
on
36
(83.
8%).
3.3.
Inst
it
ut
io
na
l
c
ha
r
act
eri
s
tics
A
to
t
a
l
of
25
f
a
c
i
l
i
t
i
e
s
we
re
i
n
c
lud
ed
in
t
h
e
s
t
ud
y.
Pu
b
l
i
c
f
a
c
i
l
i
t
i
e
s
a
c
c
ou
n
t
ed
fo
r
68
%
o
f
t
he
f
a
c
i
l
i
t
i
e
s
wh
i
l
e
f
a
i
th
b
as
ed
and
pr
i
va
t
e
in
s
t
i
tu
t
i
on
s
a
cc
ou
n
ted
fo
r
24
%
a
nd
8
%
r
e
sp
ec
t
i
v
e
l
y.
M
aj
or
i
t
y
of
th
e
f
a
c
i
l
i
t
i
e
s
,
1
2
(4
8
%)
w
er
e
l
eve
l
II
I
(h
ea
l
t
h
ce
n
tr
e
s)
,
10
(4
0%
)
l
ev
e
l
II
(d
is
p
en
sa
r
i
es
)
and
3
(1
2
%)
l
eve
l
IV
(d
i
s
t
r
i
c
t
and
s
ub
d
i
s
t
r
i
c
t
ho
sp
i
t
a
l
s)
.
D
iag
no
s
t
i
c
an
d
tr
e
a
tm
e
n
t
s
e
r
v
i
c
es
w
e
r
e
o
ff
er
ed
i
n
2
1
(8
4%
)
of
t
h
e
f
a
c
i
l
i
t
i
e
s
wh
i
l
e
4
(1
6
%)
of
fe
r
ed
on
l
y
t
r
e
at
m
e
n
t
s
er
v
i
ce
s.
A
m
on
g
t
h
e
i
n
s
t
i
t
u
t
ion
s
of
f
er
i
ng
d
i
agn
o
si
s
,
2
0
(9
5%
)
in
d
i
ca
t
ed
t
h
a
t
t
he
i
r
l
ab
s
t
a
f
f
h
ad
b
e
en
t
r
a
i
n
ed
sp
e
c
if
i
c
a
l
l
y
o
n
T
B
d
iag
no
s
i
s
p
ro
to
co
l
s.
Av
a
i
l
ab
i
l
i
t
y
o
f
a
s
p
ec
i
f
i
c
T
B
c
lin
i
c
ro
o
m
w
a
s
r
epo
r
ted
i
n
21
(8
4
%)
of
t
he
f
ac
i
l
i
t
i
e
s
w
i
th
21
(8
0%
)
i
nd
i
c
a
t
in
g
t
ha
t
sp
e
c
ific
H
CW
s
h
ad
b
ee
n
a
l
lo
c
a
t
ed
t
o
of
fe
r
T
B
se
rv
ice
s
.
A
fu
r
t
he
r
21
(
80
%
)
f
ac
i
l
i
t
i
e
s
i
nd
i
c
a
t
ed
t
h
a
t
H
CW
s
h
ad
been
t
r
a
i
ne
d
o
n
T
B
m
a
n
a
g
e
m
e
n
t
.
C
on
t
i
nu
ou
s
cou
n
s
e
l
l
in
g
w
a
s
o
f
f
er
ed
t
o
p
a
t
ien
t
s
i
n
2
3
(9
6%
)
of
t
he
s
e
i
ns
t
i
t
u
t
i
on
s
t
h
ro
ug
ho
u
t
the
t
r
e
a
t
m
e
n
t
pe
r
io
d
wh
i
l
e
19
(
79
%
)
r
epo
rt
i
n
g
th
a
t
th
e
y
h
ad
eno
ug
h
H
CW
s
to
o
ff
er
DO
T
s
up
po
r
t
t
o
p
a
t
ien
t
s.
3.4.
So
ci
o
dem
og
raph
ic
,
p
at
ie
nt
an
d
insti
tu
ti
onal
ch
ar
act
eri
s
tics
as
so
ci
ati
on
w
it
h
tre
atment
interrup
tion
Highest
le
vel
of
ed
ucati
on
r
eported
by
the
pa
rtic
ipants
s
howe
d
sta
ti
sti
cal
ly
sign
ific
ant
associat
io
n
with
treat
m
ent
interr
upti
on,
χ
2
(1,
24
3)
=
4.523,
(
p<0.033).
C
onti
nu
e
d
us
e
of
al
c
ohol
durin
g
tre
atm
ent
disp
la
ye
d
a
st
at
ist
ic
ally
sign
ific
ant
associat
ion
,
χ
2
(
1,
N=
95)
=5
.73
2,
(
p<
0.017
).
Am
on
gs
t
the
i
ns
ti
tuti
on
al
var
ia
bles,
sta
ti
sti
cal
ly
sign
ific
ant
associat
io
ns
we
re
obser
ved
betwee
n
per
cei
ved
a
vaila
bili
ty
of
adeq
uat
e
HC
W
s
to
offer
DOT
s
uppo
rt,
χ
2
(1,
275)
=8
.0005,
(
p<0.00
5
a
nd
natu
re
of
facil
it
y
offer
i
ng
treat
m
ent,
χ
2
(1,
92)
=4
.03
50,
(
p<0.04
5).
The
oth
e
r
so
ci
o
-
dem
og
ra
phic
,
pa
ti
ent
and
in
sti
tuti
on
al
va
riables
di
d
not
exh
i
bi
t
sta
ti
sti
cally
s
i
gn
i
ficant
asso
ci
at
ion
s
with
treatm
ent
in
te
rr
upti
on.
T
he
so
ci
o
-
dem
og
ra
phic
,
patie
nt
a
nd
insti
tuti
on
al
char
act
erist
ic
s
that
that
sh
owed
sta
ti
sti
cal
ly
sign
ific
ant
associat
ion
with
TB
treatm
ent
interr
up
ti
on
ar
e shown i
n
T
a
bl
e 2
.
Table
2
.
Assoc
ia
ti
on
Be
twee
n Soci
o
-
dem
og
r
aph
ic
,
Pati
e
nt a
nd
I
ns
ti
tuti
onal
Characte
risti
cs
with
TB
Treatm
ent I
nter
r
up
ti
on
Ch
arac
teristics
TB T
reat
m
en
t Out
co
m
e
χ
2
(P
-
v
alu
e)
Nu
m
b
e
r
c
en
so
red
Tr
eat
m
en
t inter
rup
tio
n
Edu
catio
n
lev
el
Pri
m
a
r
y
9
5
(96
.14
%)
3
(3.0
6
%)
4
.5(0
.03
3
)
Seco
n
d
ary
1
3
0
(89
.6
6
)
1
4
(10
.34
%)
Alco
h
o
l on
tr
eat
m
en
t
Yes
8
(72
.73
%)
3
(27
.27
%)
5
.7(0
.01
7
)
No
7
9
(94
.05
%)
5
(5.9
5
%)
Su
f
f
icien
t staff
f
o
r
DOT
Yes
2
2
7
(95
.8%
)
1
0
(
4
.2%
)
8
.0(0
.00
5
)
No
3
2
(84
.2%
)
6
(14
.8%
)
Natu
re
o
f
f
acility
Pu
b
lic
1
8
4
(91
.5%
)
1
7
(8.5
%)
4
.0(0
.04
5
)
No
n
-
p
u
b
lic
8
9
(97
.8%
)
2
(2.2
%)
Evaluation Warning : The document was created with Spire.PDF for Python.
IJPHS
IS
S
N:
22
52
-
8806
Timin
g and De
te
rminan
ts
of T
ub
e
rcul
os
is Tr
eatme
nt Interr
up
ti
on i
n
N
airo
bi
....
(
Viole
t
Jepchu
mba)
207
3.5.
Surv
i
va
l
anal
ys
is
Ou
t
of
the
291
patie
nts,
19
(
6.5%)
inter
r
up
te
d
t
heir
tre
atm
ent.
The
m
edian
ti
m
e
t
o
treat
m
ent
interr
up
ti
on
w
as
56
[
95%
CI
,
36
-
105]
days
.
Nine
(
47%)
of
the
treat
m
e
nt
interr
up
ti
ons
occurre
d
wit
hin
the
first two
m
on
ths of treatm
ent
. K
apla
n
-
Me
ie
r
f
ai
lu
re esti
m
ate
accor
ding to
highest l
evel of ed
uc
at
ion
(
l
og r
an
k
te
st;
P<0.
018)
and
al
co
hol
use
du
ri
ng
treat
m
ent
(log
ra
nk
te
st;
P<0.
01
7)
disp
la
ye
d
a
sign
i
ficant
dif
fe
ren
c
e
sh
ow
n
in
F
i
gure
1.
O
f
the
in
sti
tuti
on
al
fact
or
s
,
pe
rceive
d
avail
abili
ty
of
adequate
HC
Ws
to
offer
D
OT
(l
og
rank
te
st;
P<0.004
)
an
d
nature
of
f
aci
li
ty
of
f
erin
g
the
treat
m
ent
(log
r
an
k
te
st;
P<0.
04
7)
,
si
m
il
arl
y
disp
la
ye
d
a
sta
ti
sti
cally
sign
ific
ant
dif
fer
e
nce s
how
n
i
n
F
igure
2.
Figure
1
.
Ka
pl
an
-
Me
ie
r
f
ai
lu
r
e estim
at
es f
or
so
ci
o
-
dem
ogra
ph
ic
a
nd
patie
nt
ch
aracte
risti
cs that s
howe
d
sign
ific
a
nt ass
ociat
ion
with
T
B
treatm
ent interr
up
ti
on
Figure
2
.
Ka
pl
an
-
Me
ie
r
f
ai
lu
r
e estim
at
es f
or
insti
tuti
on
al
c
ha
racteri
sti
cs
th
at
showe
d
si
gnific
ant ass
ociat
ion
with
TB
t
reatm
ent inter
ruptio
n
3.6.
Risk
of t
re
atment in
terru
p
tion
3.6.1.
So
ci
o
-
dem
ogr
ap
hic,
P
at
ie
n
t Ch
ara
c
teristic
s
and
Risk
of
Tr
eat
me
nt
Inte
rrupti
on
Durin
g
the
ent
ire
treat
m
ent
per
io
d,
a
3.3
8
i
ncr
ease
in
risk
for
treat
m
ent
interr
up
ti
on
w
as
obser
ve
d
a
m
on
gst
patie
nts
with
seco
ndary
le
vel
e
du
cat
ion
wh
e
n
c
om
par
ed
to
th
os
e
with
pr
im
ary
le
vel
ed
uc
at
ion
.
Si
m
il
arly
,
con
ti
nu
e
d
us
e
of
a
lc
ohol
durin
g
t
reatm
ent
incre
ased
ris
k
f
or
treat
m
ent
interr
up
ti
on
by
4.04
fo
ld
com
par
ed
to
th
os
e
w
ho
did
not
con
s
um
e
al
c
ohol
durin
g
tre
atm
ent
.
The
propo
rtion
of
pat
ie
nts
that
interr
up
te
d
TB
treat
m
ent
durin
g
the
f
ollow
-
up
pe
rio
d
cl
us
te
red
acco
rd
i
ng
t
o
highe
st
le
vel
of
ed
ucati
on
an
d
al
cohol
consum
ption
duri
ng treat
m
ent
is sho
wn in
F
igure
3
.
0
.
0
0
0
.
2
5
0
.
5
0
0
.
7
5
1
.
0
0
Pro
b
a
b
i
l
i
t
y
o
f
t
re
a
t
me
n
t
i
n
t
e
rru
p
t
i
o
n
0
2
4
6
8
10
Su
rv
i
v
a
l
t
i
m
e
i
n
m
o
n
t
h
s
Pri
m
a
ry
Se
co
n
d
a
ry
Po
st
se
co
n
d
a
ry
Ka
p
l
a
n
-Me
i
e
r
f
a
i
l
u
re
e
st
i
m
a
t
e
s
f
o
r
h
i
g
h
e
st
l
e
v
e
l
o
f
e
d
u
ca
t
i
o
n
0
.
0
0
0
.
2
5
0
.
5
0
0
.
7
5
1
.
0
0
Pro
b
a
b
i
l
i
t
y
o
f
t
re
a
t
me
n
t
i
n
t
e
rru
p
t
i
o
n
0
2
4
6
8
Su
rv
i
v
a
l
t
i
m
e
i
n
m
o
n
t
h
s
N
o
a
l
co
h
o
l
U
se
d
a
l
co
h
o
l
Ka
p
l
a
n
-Me
i
e
r
f
a
i
l
u
re
e
st
i
m
a
t
e
s
f
o
r
a
l
co
h
o
l
u
se
d
u
ri
n
g
t
re
a
t
me
n
t
0
.
0
0
0
.
2
5
0
.
5
0
0
.
7
5
1
.
0
0
Pro
b
a
b
i
l
i
t
y
o
f
t
re
a
t
me
n
t
i
n
t
e
rru
p
t
i
o
n
0
2
4
6
8
10
Su
rv
i
v
a
l
t
i
m
e
i
n
m
o
n
t
h
s
I
n
a
d
e
q
u
a
t
e
H
C
W
s
Ad
e
q
u
a
t
e
H
C
W
s
Ka
p
l
a
n
-Me
i
e
r
f
a
i
l
u
re
e
st
i
m
a
t
e
s
f
o
r
a
d
e
q
u
a
cy
o
f
H
C
W
s
0
.
0
0
0
.
2
5
0
.
5
0
0
.
7
5
1
.
0
0
Pro
b
a
b
i
l
i
t
y
o
f
t
re
a
t
me
n
t
i
n
t
e
rru
p
t
i
o
n
0
2
4
6
8
10
Su
rv
i
v
a
l
t
i
m
e
i
n
m
o
n
t
h
s
Pu
b
l
i
c
N
o
n
p
u
b
l
i
c
Ka
p
l
a
n
-Me
i
e
r
f
a
i
l
u
re
e
st
i
m
a
t
e
s
f
o
r
n
a
t
u
re
o
f
f
a
ci
l
i
t
y
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2252
-
8806
IJPHS
V
ol.
6
, No
.
3
,
Septem
ber
201
7
:
203
–
21
2
208
Figure
3
.
Pro
portio
n of
treat
m
ent interr
upti
on
du
rin
g
the
foll
ow
-
up p
e
rio
d gro
up
e
d
acc
ordi
ng to hig
hes
t l
evel
of ed
ucati
on at
ta
ined
a
nd alco
ho
l c
ons
um
pti
on du
rin
g
TB t
reatm
ent
3.6.2.
Inst
it
ut
io
na
l
f
actors
and ri
s
k for
treatme
nt
in
terrup
tio
n
Faci
li
ti
es
per
cei
ved
to
ha
ve
in
adequate
sta
ff
to
offer
D
OT
s
upport
sho
wed
a
3.
69
inc
reas
e
in
risk
for
treatm
ent
inter
ruptio
n
wh
e
n
com
par
ed
t
o
f
aci
li
ti
es
that
wer
e
pe
rceiv
e
to
ha
ve
a
dequat
e
sta
ff.
A
ddit
ion
al
ly
,
public
facil
it
ie
s
ha
d
a
3.83
increase
in
risk
for
treat
m
ent
interr
up
ti
on
w
hen
com
par
ed
to
patie
nts
t
rea
te
d
in
pr
i
vate
or
fait
h
based
facil
it
ie
s.
The
pro
porti
on
of
patie
nts
that
interr
up
te
d
TB
treatm
ent
durin
g
the
f
ollow
-
up
per
i
od
cl
us
te
re
d
ac
c
ordin
g
to
per
cei
ved
a
de
quacy
of
HC
Ws
in
the
treat
m
e
nt
center
an
d
na
ture
of
t
he
fa
ci
li
t
y
offer
i
ng TB tre
atm
ent
is show
n
in
F
ig
ure
4.
0
.
0
2
.
0
4
.
0
6
.
0
8
.1
Pro
p
o
rt
i
o
n
o
f
t
re
a
t
me
n
t
i
n
t
e
rru
p
t
i
o
n
s
0
2
4
6
8
10
Su
rv
i
v
a
l
t
i
m
e
i
n
m
o
n
t
h
s
Pri
m
a
ry
l
e
v
e
l
Se
co
n
d
a
ry
l
e
v
e
l
H
i
g
h
e
st
l
e
ve
l
o
f
e
d
u
ca
t
i
o
n
0
.1
.2
.3
Pro
p
o
rt
i
o
n
o
f
t
re
a
t
me
n
t
i
n
t
e
rru
p
t
i
o
n
s
0
2
4
6
8
10
Su
rv
i
v
a
l
t
i
m
e
i
n
m
o
n
t
h
s
N
o
a
l
co
h
o
l
d
u
ri
n
g
t
re
a
t
me
n
t
Al
co
h
o
l
d
u
ri
n
g
t
re
a
t
me
n
t
A
l
co
h
o
l
co
n
su
m
p
t
i
o
n
Evaluation Warning : The document was created with Spire.PDF for Python.
IJPHS
IS
S
N:
22
52
-
8806
Timin
g and De
te
rminan
ts
of T
ub
e
rcul
os
is Tr
eatme
nt Interr
up
ti
on i
n
N
airo
bi
....
(
Viole
t
Jepchu
mba)
209
Figure
4
.
Pro
portio
n of
treat
m
ent interr
upti
on
du
rin
g
the
foll
ow
-
up grou
ped acco
r
ding
to p
e
rceive
d
a
de
qu
acy
of H
C
Ws
t
o offer D
OT
s
uppo
rt and
natu
re
of f
aci
li
ty
o
ff
e
rin
g
TB t
reatm
ent
3.6.3.
Cox pro
po
r
tio
na
l h
az
ard
Alco
ho
l
us
e
duri
ng
t
reatm
ent,
HR=
4.821
(95%
C
I
1.152
-
20.
178)
an
d
perce
ive
d
a
va
il
abili
ty
of
adequate
HC
Ws
t
o
offer
D
OT
s
uppo
rt
H
R
=
.
253
(
95%
CI
.09
2
-
.
697)
s
howe
d
si
gn
i
ficant
ass
ociat
ion
wit
h
treatm
ent o
utc
om
e
on
un
i
var
i
at
e analy
sis
show
n
i
n
T
a
ble
3.
T
able
3
.
U
niv
a
riat
e
Cox P
rop
or
ti
onal
Hazar
d An
al
ysi
s
of
F
act
or
s
Associ
at
ed
with TB
Treat
m
ent I
nter
r
up
ti
on
Variable
Un
iv
ariate
Hazar
d
Ratio
(95
% CI)
Z
-
Stat
p
>
|
z
|
Value
Pu
b
lic vs
.
n
o
n
-
p
u
b
lic
.25
3
(
.05
5
-
1
.09
7
)
-
1
.84
0
.06
6
Ad
eq
u
ate vs
I
n
ad
eq
u
ate HCW
s
.25
3
(.
0
9
2
-
.69
7
)
-
2
.66
0
.00
8
Pri
m
a
r
y
vs
Secon
d
ary
3
.42
(.
9
9
-
1
1
.81
4
)
1
.94
0
.05
2
Alco
h
o
l us
e du
ring
treat
m
en
t
4
.82
(1.1
5
-
2
0
.18
)
2
.15
0
.03
1
Natu
re
of
fac
il
ity,
adequac
y
of
HC
Ws
and
highest
le
vel
of
e
duca
ti
on
we
re
incl
ud
e
d
in
t
he
m
ul
ti
var
ia
te
analy
sis.
Alc
ohol
us
e
du
rin
g
treat
m
ent
was
excl
ud
e
d
from
the
m
od
el
due
to
lo
w
patie
nts
cons
um
ing
al
c
ohol
durin
g
t
r
eatm
ent.
The
three
facto
rs
s
howe
d
sta
ti
sti
cal
ly
sign
ific
a
nt
ass
ociat
ion
s
with
treatm
ent d
efa
ult, χ
2
(
3,2
26) =
17.37, (
p<
0.0
006)
s
how
n
in
T
able
4
.
0
.
0
5
.1
.
1
5
.2
Pro
p
o
rt
i
o
n
o
f
t
re
a
t
me
n
t
i
n
t
e
rru
p
t
i
o
n
s
0
2
4
6
8
10
Su
rv
i
v
a
l
t
i
m
e
i
n
m
o
n
t
h
s
I
n
a
d
e
q
u
a
t
e
H
C
W
s
Ad
e
q
u
a
t
e
H
C
W
s
P
e
r
ce
i
v
e
d
a
d
e
q
u
a
cy
o
f
H
C
W
s
0
.
0
2
.
0
4
.
0
6
.
0
8
Pro
p
o
rt
i
o
n
o
f
t
re
a
t
me
n
t
i
n
t
e
rru
p
t
i
o
n
s
0
2
4
6
8
10
Su
rv
i
v
a
l
t
i
m
e
i
n
m
o
n
t
h
s
Pu
b
l
i
c
f
a
ci
l
i
t
i
e
s
Pri
v
a
t
e
a
n
d
f
a
i
t
h
b
a
se
d
N
a
t
u
r
e
o
f
f
a
ci
l
i
t
y
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2252
-
8806
IJPHS
V
ol.
6
, No
.
3
,
Septem
ber
201
7
:
203
–
21
2
210
Table
4
.
M
ulti
var
ia
te
Co
x Pr
oport
ion
al
Haz
ard A
naly
sis
of
Fact
ors
Assoc
ia
te
d
with
TB
Treatm
ent I
nterrup
ti
on
Variable
Multiv
ariate
Ad
ju
sted
Haza
rd R
atio
Z
-
Stat
p
>
|
z
|
Value
(95
% CI)
Pu
b
lic vs
.
n
o
n
-
p
u
b
lic
.21
0
(
.04
6
-
.95
2
)
-
2
.02
0
.04
3
Ad
eq
u
ate vs
I
n
ad
eq
u
ate HCW
s
.19
5
(
.06
8
-
.0.5
6
)
-
3
.03
0
.00
2
Pri
m
a
r
y
vs
Secon
d
ary
5
.28
(
1
.18
-
2
3
.59
)
2
.18
0
.02
9
4.
DISCU
SSI
ON
4.1.
Tr
eat
me
nt
Inte
rrupti
on
In
this
stu
dy
6.5%
of
the
ne
w
SM+
TB
patie
nts
interrupted
their
treat
m
ent.
In
Ke
nya,
treatm
ent
interr
up
ti
on
ra
te
s
of
4.5
%
and
8.5%
ha
ve
been
re
port
ed
am
on
gs
t
ne
w
an
d
retrea
t
m
ent
T
B
pat
ie
nts’
resp
ect
ively
[
10]
.
Accor
ding
to
DLTL
D
20
15
re
port,
the
country’
s
treat
m
ent
interrupti
on
rates
w
ere
4.3%
and
3.4
%
f
or
the
two
groups
’
res
pecti
vely
[2
]
.
Ot
her
st
udie
s
rep
ort
ed
tr
eatm
ent
interru
ptio
n
rates
of
7.
2
%
a
m
on
g
PTB
pa
ti
ents
in
S
ou
t
h
Africa
[
11]
an
d
8.9%
in
Ethi
op
ia
[12].
Wh
en
al
l
f
or
m
s
of
TB
are
c
onsid
ered,
repor
te
d
treat
m
ent
interrupt
ion
rates
i
nclu
ded
11.
1%
[
13]
in
Ethio
pia,
7.4
%
in
I
nd
ia
[1
4]
a
nd
11
.5
%
in
Kuwait
[
15
]
.
Ther
e
is
a
va
riat
ion
i
n
trea
t
m
ent
interrup
ti
on
r
at
es
bet
ween
an
d
within
c
ount
ries.
These
var
ia
ti
ons
co
ul
d
be
at
trib
uted
to
the
sta
tus
of
TB
co
ntr
ol
pro
gr
am
s
within
the
stud
y
re
gions
in
-
c
ount
ry
and
betwee
n
c
ountr
ie
s.
4.2.
Ti
me to
Tr
e
at
ment
I
nt
err
up
tion
The
m
edian
tim
e to d
efau
lt
w
as 56 d
ay
s wi
th
m
ajo
rity
o
f
the p
at
ie
nts d
ef
aulti
ng
du
rin
g
the trans
it
ory
ph
a
se
bet
ween
intensive
a
nd
con
ti
nuat
io
n
ph
a
se
of
treat
m
ent.
The
res
ults
wer
e
co
nsi
ste
nt
with
re
porte
d
increase
in
haz
ard
for
treat
m
e
nt
interr
upti
on
durin
g
inten
siv
e
ph
ase
i
n
Ke
ny
a
[10].
Sim
i
lar
res
ults
hav
e
bee
n
ob
s
er
ved
in
I
ndia
w
he
re
up
t
o
40%
of
de
fa
ult
cases
occ
urred
duri
ng
the
sam
e
per
iod
[
14]
.
I
n
K
uw
ai
t
56%
of
treatm
ent
interrupti
on
we
re
obser
ve
d
withi
n
the
first
tw
o
m
on
th
s
of
trea
t
m
ent
[1
5]
.
In
Mold
ov
a
,
the
m
edian
tim
e
to
def
ault
was
110
days
an
d
the
highe
st
risk
was
ob
serv
e
d
i
n
the
m
on
th
i
m
m
edi
at
el
y
after
intensiv
e
ph
a
se
of
tre
at
m
ent
[16]
.
T
he
re
is
need
to
a
ddress
i
nd
i
vidual
an
d
healt
h
syst
e
m
chan
ge
s
occ
urrin
g
durin
g
t
he
transiti
on
ph
a
s
e
of
treat
m
ent
in
order
to
in
s
ti
tute
app
r
opri
at
e
m
easur
es
that
would
reduce
loss
to
f
ol
low
up
durin
g
this
peri
od
.
T
hese
m
ea
su
res
s
houl
d
be
pu
t
in
place
im
m
ediat
ely
the
patie
nts
beg
i
n
TB
treatm
ent
as
the
risk f
or
i
nterru
ption i
s
highest wit
hi
n t
he first
two m
on
ths.
4.3.
R
isk
F
ac
to
r
s
4.3.1.
Highes
t
Le
vel
of
Educ
at
i
on
Ther
e
was
a
si
gn
i
ficant
ass
oc
ia
ti
on
bet
ween
highest
le
vel
of
e
ducat
io
n
at
ta
ined
by
the
pa
ti
ents
an
d
TB
treatm
ent
interr
up
ti
on
.
A
3.3
8
inc
rea
se
in
risk
for
TB
treatm
ent
interr
up
ti
on
w
as
ob
ser
ved
a
m
on
gs
t
patie
nts
with
s
econda
ry
le
vel
edu
cat
io
n
w
he
n
com
par
ed
t
o
tho
s
e
with
pri
m
ary
le
vel
e
du
cat
io
n.
T
he
fin
dings
in
this
st
ud
y
t
her
e
fore
pr
e
se
nt
a
po
s
sible
presence
of
a
hi
gh
risk
gr
oup
previ
ou
sly
not
consi
der
e
d.
Fu
rt
her
inv
est
igati
on
i
nto
at
ti
tud
e
s
a
nd
healt
h
s
eek
ing
be
hav
i
or
s
i
n
this
cat
e
gory
of
patie
nts
ne
ed
to
be
e
valu
at
ed
to
unde
rstan
d
the
reason
f
or
i
ncrea
sed risk
in
t
hi
s g
r
oup.
4.3.2.
Alcohol
u
se
D
uring Tre
atm
ent
A
sta
ti
sti
cal
ly
sign
ific
a
nt
ass
ociat
ion
was
obser
ve
d
betwe
en
c
on
ti
nu
e
d
al
cohol
us
e
a
nd
treat
m
ent
ou
tc
om
e
.
Pati
ents
who
c
onti
nu
e
d
al
co
ho
l
us
e
duri
ng
tre
atm
ent
had
a
4.04
inc
rease
in
ris
k
for
t
re
atm
ent
interr
up
ti
on
co
m
par
ed
to
t
ho
s
e
w
ho
di
d
no
t
consum
e
al
cohol.
New
cases
with
al
co
holi
sm
and
dru
g
a
ddic
ti
on
hav
e
bee
n
s
how
n
t
o
ha
ve
an
inc
reas
e
d
hazar
d
of
de
f
aulti
ng
[16
],
[
17]
.
I
n
this
stu
dy
m
ajo
rity
of
stu
dy
par
ti
ci
pa
nts
wi
th
a
histor
y
of
al
cohol
us
e,
s
topped
al
c
ohol
con
s
um
ption
durin
g
TB
trea
t
m
ent.
Alcohol
us
e
i
m
pairs
m
enta
l
capa
bili
ty
and
this
inter
fere
s
with
decisi
on
m
aking
ca
pacit
y
inclu
din
g
ad
he
ren
ce
to
T
B
treatm
ent.
The
re
has
bee
n
a
c
al
l
to
act
ion
to
address
al
c
ohol
us
e
a
nd
ab
use
to
achie
ve
optim
al
resu
lt
s
in
TB
pr
e
ve
nt
an
d
tre
atm
ent
[1
8].
P
at
ie
nts
who
c
onti
nu
e
us
e
of
a
lc
ohol
duri
ng
TB
treatm
ent
sh
ould
be
c
onsi
der
e
d
high
risk f
or
t
r
eatm
ent
interr
upti
on.
4.3.3.
Availabil
it
y
of
Ad
e
qu
at
e
He
alth C
are
Wor
kers
Perceive
d
ina
de
qu
at
e a
vaila
bili
ty
o
f
HC
W
s to
offe
r
D
OT
w
as iden
ti
fied
as
a r
isk f
act
or fo
r
treat
m
ent
interr
up
ti
on
.
T
her
e
was
a
3.6
9
f
inc
rease
in
risk
f
or
treat
m
ent
inter
r
up
ti
on
am
ong
patie
nts
treat
e
d
in
f
aci
li
ti
es
that
wer
e
per
c
ei
ved
to
ha
ve
inade
quat
e
HC
Ws
to
offer
D
OT.
A
f
un
ct
io
nal
healt
h
syst
e
m
req
uires
suffici
ent
healt
h
wor
ker
s
,
eq
uitably
distrib
uted
t
o
im
pr
ov
e
co
ve
rag
e
a
nd
acce
ssibil
it
y.
The
nee
d
f
or
a
dequate
heal
thcare
workers
with
the
ri
gh
t
sk
il
ls
can
no
t
be
ov
erem
ph
asi
zed
as
a
requirem
ent
to
achie
vin
g
healt
h
go
al
s
in
a
Evaluation Warning : The document was created with Spire.PDF for Python.
IJPHS
IS
S
N:
22
52
-
8806
Timin
g and De
te
rminan
ts
of T
ub
e
rcul
os
is Tr
eatme
nt Interr
up
ti
on i
n
N
airo
bi
....
(
Viole
t
Jepchu
mba)
211
popula
ti
on
.
T
her
e
is
lim
it
e
d
li
te
ratur
e
publis
hed
per
ta
ining
t
o
HC
W
sta
ff
i
ng
le
vels
an
d
TB
treatm
ent
ou
tc
om
es.
The
re
is
a
n
urge
nt
nee
d
to
ri
gor
ou
sly
e
valuate
the
im
po
rta
nc
e
of
a
dequate
sta
ff
in
g
t
o
pro
vid
e
evide
nce
in
de
sign
i
ng
op
ti
m
a
l
TB
treatm
ent
gu
i
delines.
T
he
resu
lt
s
are
s
uppo
rtive
of
the
hig
hli
gh
te
d
ne
ed
f
or
strat
egies to
tra
ns
f
or
m
h
eal
th
workf
or
ce
cap
abili
ti
es in order to
achie
ve
th
e target t
o
e
nd
TB [
19
]
.
4.3.4.
Nature
of
F
ac
il
ity
Tu
ber
c
ulo
sis
tr
eatm
ent
is
offe
red
in
pu
blic
and
no
n
-
public
facil
it
ie
s
in
Ke
nya.
T
he
nat
ure
of
facil
it
y
offer
i
ng
treat
m
ent
sh
owe
d
a
sta
ti
sti
cally
sign
ific
a
nt
ass
ociat
i
on
to
t
re
atm
ent
interrupti
on.
Pati
ents
who
receive
d
treat
m
ent
in
public
facil
it
ie
s
had
a
3.83
i
ncr
ease
in
ris
k
f
or
t
reatm
ent
interr
up
ti
on
w
hen
c
om
par
ed
t
o
tho
se
w
ho
re
cei
ved
treat
m
ent
in
non
-
pu
blic
facil
it
ie
s.
Si
m
il
ar
resu
lt
s
wer
e
repo
rted
in
Nige
ria
wh
e
re
receivin
g
treat
m
ent
at
a
pu
blic
facil
it
y
was
a
pr
e
dictor
of
un
s
ucces
sf
ul
treatm
e
nt
ou
tc
om
es
[20].
Wh
il
e
ther
e
has
be
en
no
ev
idence
to
sho
w
that
there
is
be
tt
er
cl
inica
l
serv
ic
es
in
non
-
public
facil
it
ies,
patie
nts
in
pri
vate
facil
it
ie
s
rep
or
t
ed
bette
r
inter
pe
rsonal
sat
isfa
ct
ion
at
tribu
te
d
to
lon
ge
r
co
nsult
at
ion
s
tim
es
and
hi
gh
e
r
ch
ances
of
receivi
ng
c
ouns
el
li
ng
[21],
[22].
I
n
K
en
ya
,
public
heal
th
facil
it
ie
s
are
known
t
o
be
well
stocke
d
with
TB
tracer
c
omm
o
diti
es
to
pro
vid
e
TB
ser
vice
s
[5]
.
The
re
is
nee
d
t
o
deter
m
ine
whet
her
patie
nt
inter
pe
rsonal
sat
isfact
ion
ov
er a
nd abo
ve
s
uffici
e
nt cli
nic
al
ser
vices
is a
risk fact
or to
T
B t
reat
m
ent interr
up
ti
on.
5.
CONCL
US
I
O
N
Elim
inati
ng
T
B
treatm
ent
interr
upti
on
is
a
key
facto
r
in
the
fig
ht
against
TB
.
P
recautio
na
ry
m
easur
es
need
to
be
put
in
place
f
or
patie
nts
w
ho
co
ntinu
e
al
co
ho
l
a
nd
dr
ug
us
e
du
rin
g
TB
treat
m
ent
to
pr
e
ve
nt
treatm
ent
interr
upti
on.
Healt
h
syst
e
m
m
anag
em
ent
chan
ge
s
occ
urrin
g
du
rin
g
th
e
transiti
on
between
intensive
an
d
con
ti
nuat
io
n
phase
of
t
reatm
ent
nee
d
t
o
be
evaluate
d
to
determ
ine
the
reason
for
in
cr
ease
d
treatm
ent
inter
ruptio
n
du
rin
g
that
pe
rio
d
of
treat
m
ent.
This
stu
dy
highli
ghts
the
nee
d
to
look
at
the
hu
m
an
resou
rce
com
po
ne
nts
in
healt
h
facil
it
ie
s
and
the
non
-
cl
inic
al
insti
tuti
on
al
var
ia
bles
that
are
co
ntributi
ng
t
o
adv
e
rse
treat
m
ent
outc
om
es.
Pati
ent
per
ce
pt
ion
a
nd
at
ti
tude
on
qual
it
y
and
treat
m
ent
app
r
oac
h,
s
hould
to
be
integrate
d
int
o al
l healt
h
facil
i
ti
es o
f
fer
in
g T
B t
reat
m
ent.
ACKN
OWLE
DGE
MENTS
The
a
uthors
woul
d
li
ke
to
than
k
Nairob
i
County
and
sp
eci
fical
ly
TB
and
Lep
r
os
y
pro
gram
le
ader
s
hip
f
or
al
lowi
ng
acc
ess
to
healt
h
f
aci
li
ti
es
to
intervie
w
patie
nts
an
d
re
view
TB
re
gisters
f
or
dat
a
colle
ct
ion
.
We
al
so
ex
press
our
gr
at
it
ude
to
al
l
healt
h
care
wor
ker
s
a
nd
TB
pat
ie
nts
f
or
pa
rtic
ipati
ng
in
the
stud
y.
REFERE
NCE
S
[1]
W
HO
,
“
Global
Tube
rcu
losis Report
2016
,
”
W
or
ld
Hea
lt
h
Organi
za
t
ion
,
2016.
[2]
D
ivi
sion o
f Lepr
os
y
,
“
Tube
rcu
lo
sis a
nd
Lung
Dis
ea
se
,”
DL
TL
D
Ken
y
a
Annual
R
epor
t
,
2015
.
[3]
W
HO
,
“
Tube
rcu
losis Fac
tshe
et
No.
104
,”
W
orld
Hea
l
th
Organ
izati
on
,
2012
.
[4]
W
HO
,
“
Im
ple
me
nti
ng
END T
B
strat
eg
y
:
THE
E
SS
ENTI
ALS
,”
W
HO
,
2015
.
[5]
“
Ken
y
a
Se
rvice
Avail
ab
il
ity
and
Rea
din
ess As
ses
sm
ent
Mapping (S
ARA
M),
”
2013.
[6]
Ken
y
a
He
al
th
Se
ct
or
,
“
Hum
an
Re
source
s Stra
t
eg
y
2014
-
2018
,
”
20
16.
[7]
W
HO
,
“
Defi
nit
i
ons a
nd
r
epor
ti
n
g
–
7
fra
m
ework
f
or
tube
r
cul
osis
,”
2013.
[8]
Division
of
Le
p
ros
y
,
“
Tube
r
culos
is
and
Lung
Disea
se,
”
DL
T
LD
Guidel
in
es
on
m
ana
gement
of
Le
pros
y
an
d
Tube
rcu
losis
,
20
13
.
[9]
W
HO
,
“
Tre
at
m
e
nt
of tuber
cul
osi
s: gui
de
li
nes
–
4th
ed
,
W
HO
,
2010
.
[10]
E.
O.
Ma
sini
,
et
al.
,
“
Us
ing
Survival
An
aly
sis
to
Ide
nti
f
y
Risk
Fa
ct
ors
for
Treat
m
ent
In
te
rrupt
ion
among
New
and
Ret
re
at
m
ent T
ub
erc
ulosis Pa
ti
en
t
s in
Ken
y
a
,
”
P
lo
s One,
vo
l
/i
ss
ue:
11
(
10
)
,
2016.
[11]
G.
Kigozi
,
e
t
al.
,
“
Fact
ors
infl
ue
nci
ng
treatment
def
ault
among
tube
rcu
losis
pati
ent
s
in
a
high
b
urde
n
province
of
South Afri
ca,”
I
nte
rnational
Jou
rnal
of
In
fecti
ou
s Diseases,
vol
.
54
,
pp
.
95
-
102
,
2017.
[12]
T.
Z
ene
b
e
and
E
.
Tefera,
“
Tube
r
cul
osis
treatment
outc
om
e
and
as
socia
t
ed
fa
ct
ors
among
sm
ea
r
-
positi
ve
pu
lmonar
y
tube
rcu
losis
patient
s
in
Afar,
Ea
stern
E
thi
opi
a:
a
retros
pec
t
i
ve
stud
y
,
”
The
Brazili
an
Journal
of
Infe
c
ti
ous
Diseases,
vol
/
issue:
20
(
6
)
,
pp
.
63
5
-
36,
2016
.
[13]
G.
Gebre
zg
abi
h
er,
et
a
l.
,
“
Treat
m
ent
Outcome
of
Tube
r
cul
osis
Pati
ent
s
und
er
Dire
ctl
y
Obs
erv
ed
Tr
ea
tm
ent
Short
Course
and
Fa
c
tors
Affec
t
ing
Outcome
in
So
uthe
rn
Et
hiop
ia
:
A
Five
-
Ye
ar
Ret
rospec
ti
ve
Stud
y
,
”
Pl
os
One
,
vol
/i
ss
ue:
11
(
2
)
,
2016.
[14]
K.
Vasudeva
n,
e
t
al
.
,
“
Sm
ea
r
Co
nver
sion,
Treat
m
ent
Outcomes
and
the
Ti
m
e
of
Defa
ult
in
Regi
s
te
red
Tub
erc
ulos
is
Pati
ent
s
on
RNTCP
DO
TS
in
Puducher
r
y
,
Southern
India,”
Journal
Of
Cli
nic
al
And
Diag
nostic
Re
search
,
vol
/i
ss
ue:
8
(
10
)
,
2014.
[15]
Q.
Zha
ng
,
e
t
al
.
,
“
Dete
rm
ina
nts
of
Defa
ul
t
fro
m
Pulm
onar
y
T
uber
cul
osis
Tr
eatm
ent
in
Kuw
ait,
”
The
Sc
ie
nt
if
i
c
World
Journal,
vol.
2014
,
2014
.
Evaluation Warning : The document was created with Spire.PDF for Python.
IS
S
N
:
2252
-
8806
IJPHS
V
ol.
6
, No
.
3
,
Septem
ber
201
7
:
203
–
21
2
212
[16]
H.
E.
J
enki
ns,
et
al
.
,
“
Risk
factor
s
and
ti
m
ing
of
def
ault
from
trea
tment
for
non
-
m
ult
idrug
-
r
esista
n
t
tube
r
cul
osis
in
Moldova,
”
The
I
nte
rnational
Jou
rnal
of
Tubercu
l
osis and
Lung Disease,
vo
l
/i
ss
ue:
17
(
3
)
,
pp
.
373
-
8
0,
2013
.
[17]
Ravi
gli
on
e
M.
and
Pozn
y
ak
V.
,
“
Ta
rge
ti
ng
har
m
ful
use
of
al
cohol
for
pre
vent
io
n
and
tre
at
m
ent
of
tube
rcu
losis:
a
ca
l
l
for
action
,
”
European
R
espiratory
Journal
,
v
ol
/i
ss
ue:
50
(
1
)
,
2
017
.
[18]
A.
Finlay
,
et
a
l
.
,
“
Pati
ent
-
and
pr
ovide
r
-
l
eve
l
risk
fac
tors
associa
t
ed
with
de
fau
l
t
f
rom
tube
rcu
losis
tre
a
tment,
Sout
h
Afric
a, 2002: a c
ase
-
cont
ro
l
stud
y
,
”
BMC Publ
i
c Healt
h,
vol
/
issue:
12
(
1
)
,
2012.
[19]
W
HO
,
“
Global
strateg
y
on
hum
an
resourc
es
for
h
ea
l
th:
W
orkforce
2030
,”
W
orld
H
ea
l
th
Organi
zati
on
,
2016
.
[20]
K.
N.
Ukw
aj
a,
e
t
al.
,
“
Profile
an
d
det
erminant
s
o
f
unsucc
essful
tube
rcu
losis
outc
o
m
e
in
rura
l
Nige
ria
:
Im
pli
c
at
ions
for
tube
r
cul
osis
cont
rol
,
”
World
Journal
of
Me
th
odology
,
v
ol
/
issue:
6
(
1
)
,
pp
.
118
-
125,
2016
.
[21]
I.
Haz
ar
ika
,
“
Role
of
private
s
ec
tor
in
prov
iding
tube
rcu
losis
ca
re
:
Evi
d
ence
f
rom
a
popula
ti
o
n
-
base
d
surve
y
in
India
,
”
Journal of
Global
Infect
i
ous Diseases,
vo
l
/i
ss
ue:
3
(
1
)
,
pp.
19
-
24,
2011
.
[22]
P.
R
.
E
liy
a
,
e
t
al
.
,
“
The
qualit
y
o
f
outpa
t
ie
nt
pri
m
ar
y
c
are
in
pu
bli
c
and
priv
at
e
sec
tors
in
Sri
Lanka
--
how
well
d
o
pat
i
ent
per
ce
pt
io
ns
m
at
ch
rea
lit
y
and
what
are
the
implicati
ons
?
”
Healt
h
Policy
a
nd
Pl
anning,
vo
l
/i
ss
ue:
30
(
1
)
,
pp.
59
-
74
,
2015
.
Evaluation Warning : The document was created with Spire.PDF for Python.