mhealth text and voice communication for monitoring people ... · anstey watkinsfi etfial bmj glob...
TRANSCRIPT
1Anstey Watkins J, et al. BMJ Glob Health 2018;3:e000543. doi:10.1136/bmjgh-2017-000543
mHealth text and voice communication for monitoring people with chronic diseases in low-resource settings: a realist review
Jocelyn Anstey Watkins,1 Jane Goudge,2 Francesc Xavier Gómez-Olivé,3 Caroline Huxley,1 Katherine Dodd,1 Frances Griffiths1,2
Research
To cite: Anstey Watkins J, Goudge J, Gómez-Olivé FX, et al. mHealth text and voice communication for monitoring people with chronic diseases in low-resource settings: a realist review. BMJ Glob Health 2018;3:e000543. doi:10.1136/bmjgh-2017-000543
Handling editor Valery Ridde
Received 30 August 2017Revised 7 February 2018Accepted 9 February 2018
1Division of Health Sciences, Warwick Medical School, University of Warwick, Coventry, UK2Centre for Health Policy, Faculty of Health Sciences, School of Public Health, University of the Witwatersrand, Johannesburg, South Africa3MRC/Wits Rural Public Health and Health Transitions Research Unit (Agincourt), School of Public Health, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa
Correspondence toDr Jocelyn Anstey Watkins; jotawatkins@ gmail. com
AbstrACtbackground Routine monitoring by patients and healthcare providers to manage chronic disease is vital, though this can be challenging in low-resourced health systems. Mobile health (mHealth) has been proposed as one way to improve management of chronic diseases. Past mHealth reviews have proposed the need for a greater understanding around how the theoretical constructs in mHealth interventions actually work. In response, we synthesised evidence from primary studies on monitoring of chronic diseases using two-way digital text or voice communication between a patient and health worker. We did this in order to understand the important considerations for the design of mHealth interventions.Method Articles retrieved were systematically screened and analysed to elicit explanations of mHealth monitoring interventions. These explanations were consolidated into programme theory and compared with existing theory and frameworks. We identified variation in outcomes to understand how context moderates the outcome.results Four articles were identified—monitoring of hypertension and HIV/AIDS from: Kenya, Pakistan, Honduras and Mexico and South Africa. Six components were found in all four interventions: reminders, patient observation of health state, motivational education/advice, provision of support communication, targeted actions and praise and encouragement. Intervention components were mapped to existing frameworks and theory. Variation in outcome identified in subgroup analysis suggests greater impact is achieved with certain patient groups, such as those with low literacy, those with stressful life events or those early in the disease trajectory. There was no other evidence in the included studies of the effect of context on the intervention and outcome.Conclusion mHealth interventions for monitoring chronic disease in low-resource settings, based on existing frameworks and theory, can be effective. A match between what the intervention provides and the needs or social factors relevant to specific patient group increases the effect. It was not possible to understand the impact of context on intervention and outcome beyond these patient-level measures as no evidence was provided in the study reports.
IntroduCtIonThe burden of chronic diseases is an esca-lating problem in low-income and middle-in-come countries (LMICs).1 The Sustain-able Development Goals state that by 2030, improving the prevention and management of chronic communicable and non-communi-cable diseases is a priority for primary care in public health systems.2 3 Chronic diseases are long-term, disruptive and often intrusive to individuals’ everyday lives.4 The management of chronic diseases is a dynamic process that varies over time, depending on the disease
Key questions
What is already known about this topic? The burden of chronic diseases is an escalating problem globally. Evidence reviews suggest mHealth interventions delivered in low-income and middle-income countries can be effective in improving health outcome for people living with chronic disease.
What are the new findings? Effective interventions using two-way digital communication between healthcare providers and patients living with long-term conditions are based on established frameworks and behaviour change theory.
Outcome is improved for patients with low literacy, stress-inducing life events and being recently diagnosed.
Contextual factors beyond those measured at patient level were not reported in the studies.
recommendations for policy mHealth interventions for improving the monitoring of chronic disease, when based on existing frameworks and theory, have potential for improving patient care and health outcome, particularly when tailored to the needs of specific patient groups.
on June 15, 2020 by guest. Protected by copyright.
http://gh.bmj.com
/B
MJ G
lob Health: first published as 10.1136/bm
jgh-2017-000543 on 6 March 2018. D
ownloaded from
2 Anstey Watkins J, et al. BMJ Glob Health 2018;3:e000543. doi:10.1136/bmjgh-2017-000543
BMJ Global Health
aetiology and physiology.5 Chronic disease involves regular self-care6 7 and routine monitoring by patients and health workers to check disease progress or regress.8 Long-term monitoring encompasses adhering to treat-ment and capturing vital signs or clinical indicators. The purpose is to improve outcomes and quality of life9 by reducing acute exacerbations and premature death10 and to maximise health.
The monitoring of chronic diseases can be chal-lenging, particularly in low-resourced health systems limited by long distances to health facilities and low staff capacity.11 In this environment, mobile health (mHealth) has been proposed as an approach to improve manage-ment of chronic diseases 12-14 including assistance with monitoring. mHealth technologies used in LMIC include portable wireless devices, including mobile phones and tablets.15 mHealth can involve one-way or two-way communication between the health worker and patient, using any digital channel that allows the users to be mobile. Vasudevan et al16 suggest that strides are being made in strengthening the global mHealth evidence base along with the key ‘best practices’ in scaling mHealth for achieving universal health coverage.
A systematic search for mHealth reviews in seven data-bases found several mHealth reviews that included aspects of monitoring such as adherence studies,17 behaviour change18 and attendance reminders.19 20 Review authors described how the effectiveness of mHealth evidence is mixed. They suggest there is a lack of understanding of why mHealth interventions for chronic disease
management should work (or not) in LMICs.17 18 20–27 One of these reviews proposed that future studies should explicitly describe the theoretical constructs that mHealth interventions are targeting.28 They argue this will make it clearer how and why the intervention is intended to work and in what circumstances.28 A stronger theoret-ical understanding is likely to strengthen the mHealth evidence base.27 However, this search revealed that there is no systematic review set in LMICs specifically focusing on one-way and two-way mHealth communication for monitoring of chronic disease.
This paper responds to the challenge identified in the systematic reviews, by examining the theoretical founda-tions and mechanisms by which mHealth interventions work (or not), to support chronic disease monitoring in LMICs.
We focus on understanding how monitoring of chronic diseases may be improved using two-way digital communi-cation between a patient and health worker (community health worker/nurse/doctor) in a low-resourced setting to potentially guide future intervention design.
MetHodthe realist method and initial searchA ‘realist review’, also known as a ‘realist synthesis’, makes sense of heterogeneous evidence about complex interventions applied in diverse contexts.29 30 It draws on substantive theory and empirical research from across multiple disciplines31 to understand how interventions
Figure 1 mHealth example of a CMOc summarised from the reviews. CMOc, Context-Mechanism-Outcome configuration; LMIC, low-income and middle-income countries.
on June 15, 2020 by guest. Protected by copyright.
http://gh.bmj.com
/B
MJ G
lob Health: first published as 10.1136/bm
jgh-2017-000543 on 6 March 2018. D
ownloaded from
Anstey Watkins J, et al. BMJ Glob Health 2018;3:e000543. doi:10.1136/bmjgh-2017-000543 3
BMJ Global Health
work and for whom. By using the methods of a realist review,32–34 we questioned what works, compared with what, how well, with what exposure, with what behav-iours, for how long, for whom, in what settings and why35? We went through the following realist stages: (1) identi-fying programme theory, how the study authors intended their interventions to work; (2) testing this programme theory against empirical evidence and established high-level theory and (3) providing guidance for future inter-vention development.
We did a preliminary systematic search for systematic reviews of mHealth. From the 10 relevant reviews iden-tified,17 18 20–27 we extracted examples of how the review authors thought the interventions worked from their included papers. We mapped these on to the layout of the realist Context-Mechanism-Outcome configuration (CMOc) and adapted the diagram designed by Dalkin36 (figure 1) using the example of mHealth monitoring. Figure 1 illustrates how mechanisms are defined as the reactions or responses to the resources available within an interaction process that leads to outcomes.37 Mecha-nisms are the responses to the intervention or patients’ resources and are contingent and conditional, only firing in particular contexts.38 The outcomes are the desired response to the resources resulting from the participants’ reasoning39 within a particular context.
Main search to identify empirical papersTo identify relevant empirical studies, we conducted the main search to systematically find papers on the moni-toring of chronic disease using two-way text message or voice call interventions in LMICs, written in English, published from January 2000 to March 2017. We limited the search to the year 2000 or later as mobile phones
were not widely available before then.28 Seven appro-priate databases and 12 search terms were used. Titles and abstracts were screened against the inclusion and exclusion criteria, as detailed in table 1. Each reference list of the final included papers in this review was checked to identify further relevant primary research studies.
data extraction from empirical papersFrom the included studies, we extracted data (table 2) on: Research design, participant sample, setting, outcome measures; intervention description, components and study duration; Intervention effect; Authors’ programme theory where stated and our interpretation of whether the study worked as intended.
Where authors did not describe how they thought their intervention would work, we worked this out from careful reading of the description of the intervention. We also referred to related study protocols, if available, or other accompanying papers such as process evaluations of the trial. JAW, FG and KD extracted data independently and then compared data to resolve any inconsistencies.
Identifying established theory to test the programme theoryWe used the CMOc in the diagram in figure 1 and applied this to each empirical study on chronic disease moni-toring found in the search.33 We developed a summary of how the interventions were intended to work according to the study authors. We then searched for relevant estab-lished high-level theory related to study intervention components and through further literature searching. As most of the mechanisms were behaviour-based, we looked mostly to psychology for the theory. We aimed to match established theory to each intervention component type (table 3) to help us reach our final CMOc. The theories
Table 1 Databases, search terms and inclusion and exclusion criteria for identifying empirical studies
Databases searched
Cochrane Register of Controlled Trials, Medline (Ovid), PubMed, Web of Science, Psych Info, Google Scholar, Knowledge for Health
Search terms Patient* AND low- and middle- income countr* OR LMIC OR developing* AND chronic disease* AND mobile phone* OR text messag* OR SMS OR voice AND monitor* OR self-monitor* OR self-manage*
Inclusion criteria Population: Patients with any chronic disease and all cadres of health workers in a public health system.Context: Low-resourced settings of any LMICs.Intervention design: The study interventions include two-way exchange of digital information necessary for monitoring initiated by the patient or health worker. The mHealth resource is either voice call or text message based mobile phone communication to improve monitoring of chronic disease.Monitoring: monitoring of side effects of medication, monitoring of physiological measures such as blood pressure or monitoring of how well the patient feels and offering relevant information and support and reminding them about future appointments or medication compliance.Study design: Empirical research evaluating the effectiveness of the intervention (any design). The study includes a description of the intervention.Outcomes: Impact/health and process outcomes (primary and secondary).
Exclusion criteria Studies set in high-income countries. Studies using only landline telephones or computer-based communication. Protocols or reports of intervention development with no published evidence outcome. Studies not published in English. Feasibility and pilot studies with no outcome for the intervention. Reviews and reviews of reviews.
LMICs, low-income and middle-income countries.
on June 15, 2020 by guest. Protected by copyright.
http://gh.bmj.com
/B
MJ G
lob Health: first published as 10.1136/bm
jgh-2017-000543 on 6 March 2018. D
ownloaded from
4 Anstey Watkins J, et al. BMJ Glob Health 2018;3:e000543. doi:10.1136/bmjgh-2017-000543
BMJ Global Health
Tab
le 2
D
ata
extr
actio
n of
inte
rven
tion
des
ign
and
effe
ct, a
nd a
utho
r’s p
rogr
amm
e th
eory
from
all
four
incl
uded
em
piri
cal s
tud
ies
on m
Hea
lth in
Ken
ya, S
outh
Afr
ica,
H
ond
uras
and
Mex
ico
and
Pak
ista
n
Aut
hor,
year
an
d c
oun
try
Res
earc
h d
esig
n, p
arti
cip
ant
sam
ple
, set
ting
, mai
n o
utco
me
mea
sure
and
se
cond
ary
out
com
eIn
terv
enti
on
des
crip
tio
n, in
terv
enti
on
com
po
nent
s an
d s
tud
y d
urat
ion
Inte
rven
tio
n ef
fect
on
pri
mar
y, s
eco
ndar
y an
d p
roce
ss o
utco
mes
and
ove
rvie
w
Aut
hors
’ pro
gra
mm
e th
eory
fo
r th
e in
terv
enti
on
(if s
tate
d) o
r p
rob
able
p
rog
ram
me
theo
ry, b
ased
on
des
crip
tio
n o
f in
terv
enti
on
(incl
udin
g li
kely
m
echa
nism
/int
erve
ntio
n fu
ncti
on)
Did
the
inte
rven
tio
n w
ork
as
inte
nded
or
not?
Lest
er e
t al
, 20
1044
: Ken
yaD
esig
n: R
CT
Sam
ple
: Pat
ient
s in
fect
ed
with
HIV
, ove
r 18
yea
rs o
ld,
initi
atin
g A
RT
for
the
first
-tim
e.
n=80
3 to
tal.
Set
ting:
One
urb
an u
nive
rsity
cl
inic
, one
urb
an n
on-
gove
rnm
enta
l org
anis
atio
n cl
inic
and
one
rur
al p
ublic
he
alth
ser
vice
clin
ic.
Mai
n ou
tcom
e m
easu
re:
Sel
f-re
por
ted
AR
T ad
here
nce
(>95
% o
f pre
scrib
ed d
oses
in
the
pas
t 30
day
s); p
lasm
a H
IV-
1 vi
ral R
NA
load
sup
pre
ssio
n (<
400
cop
ies
per
mL)
.S
econ
dar
y ou
tcom
e: R
ate
of a
ttrit
ion
(not
hav
ing
a fin
al
visi
t at
12
mon
ths)
and
rat
es o
f se
vera
l cat
egor
ies
of a
ttrit
ion
(mor
talit
y, w
ithd
raw
al fr
om t
he
stud
y, t
rans
fer
to n
on-s
tud
y cl
inic
s an
d lo
ss t
o fo
llow
-up
w
ithou
t id
entifi
able
cau
se).
Nur
se o
r cl
inic
al o
ffice
r se
nds
the
pat
ient
s a
text
mes
sage
: ‘H
ow a
re y
ou?’
eac
h M
ond
ay
mor
ning
. Pat
ient
s as
ked
to
resp
ond
with
in
48 h
ours
with
: ‘D
oing
wel
l’ or
‘Hav
e a
pro
ble
m’.
Nur
se/c
linic
al o
ffice
r p
hone
s th
e p
atie
nt
who
has
pro
ble
ms
or if
the
re is
no
resp
onse
. P
atie
nts
rece
ived
inte
rven
tion
trai
ning
whe
n re
crui
ted
. Tex
t m
essa
ge s
ent
usin
g m
ultip
le
reci
pie
nt b
ulk
mes
sagi
ng.
Inte
rven
tion
com
pon
ents
:
P
rovi
sion
of s
upp
ort
com
mun
icat
ion
Ob
serv
atio
n
R
emin
der
sS
tud
y d
urat
ion:
12
mon
ths
(bas
elin
e, 6
and
12
mon
ths)
Imp
act
outc
ome:
Sel
f-re
por
ted
ad
here
nce:
in
terv
entio
n gr
oup
: 168
(62%
); co
ntro
l gro
up:
132
(50%
) RR
0.8
1; 9
5% C
I (0.
69 t
o 0.
94);
p=
0·00
6V
iral s
upp
r ess
ion:
-in
terv
entio
n gr
oup
156
(5
7%);
cont
rol g
roup
128
(48%
); R
R 0
.85;
95
% C
I (0.
72 t
o 0.
99);
p=
0.04
.S
econ
dar
y ou
tcom
e: N
o si
gnifi
cant
as
soci
atio
ns w
ith t
he in
terv
entio
n.In
terv
entio
n gr
oup
: 19%
mis
sing
, 9%
m
orta
lity,
3%
with
dra
wal
, 6%
loss
to
follo
w-u
p
and
1%
tra
nsfe
rred
out
.C
ontr
ol g
roup
: 23%
mis
sing
, 11%
mor
talit
y,
1% w
ithd
raw
al, 1
0% lo
ss t
o fo
llow
-up
and
0%
tra
nsfe
r ou
t (to
a d
iffer
ent
clin
ic n
ot in
the
st
udy)
.P
roce
ss o
utco
mes
: At
the
end
of t
he s
tud
y,
191
of 1
94 p
atie
nts
in t
he in
terv
entio
n gr
oup
re
por
ted
the
y w
ould
like
the
tex
t m
essa
ge
pro
gram
me
to c
ontin
ue, o
f who
m 1
88 (9
8%)
said
the
y w
ould
rec
omm
end
it t
o a
frie
nd. I
n th
e fo
cus
grou
p s
essi
ons,
man
y p
atie
nts
in
the
inte
rven
tion
grou
p a
lso
rep
orte
d t
hat
they
th
ough
t th
e te
xt m
essa
ge s
upp
ort
serv
ice
was
va
luab
le.
Out
com
e ov
ervi
ew: S
igni
fican
t ch
ange
in s
elf-
rep
orte
d a
dhe
renc
e to
AR
T an
d s
upp
ress
ion
of v
iral l
oad
s b
etw
een
grou
ps.
Mal
e/ur
ban
re
sid
ence
favo
ured
ad
here
nce
com
par
ed w
ith
the
cont
rol g
roup
.
The
text
mes
sage
act
ed a
s an
ind
irect
re
min
der
to
the
pat
ient
s to
tak
e A
RT
and
p
rovi
ded
sup
por
t (p
erce
ived
sup
por
t w
as
valu
able
).Th
e p
atie
nts
know
the
y w
ill b
e fo
llow
ed u
p
if th
ey d
o no
t re
spon
d, s
o th
e d
esig
n of
the
in
terv
entio
n ac
ts a
s an
ince
ntiv
e to
res
pon
d
othe
rwis
e th
e nu
rse
or c
linic
al o
ffice
r w
ill
cont
act
them
.B
y re
ceiv
ing
a w
eekl
y m
essa
ge t
hat
asks
ho
w t
hey
are,
the
pat
ient
is m
ade
to fe
el
care
d fo
r d
urin
g th
e fir
st p
erio
d t
hey
are
taki
ng H
IV t
reat
men
t.
Yes—
the
feel
ing
of
bei
ng s
upp
ort
ed v
ia
dig
ital
co
mm
unic
atio
n w
as im
po
rtan
t. T
he t
ext
mes
sage
inte
rven
tion
was
w
ell r
ecei
ved
by
pat
ient
s,
man
y of
who
m r
epor
ted
th
at t
hey
felt
‘like
som
eone
ca
res’
. Pat
ient
s ha
d t
o re
ply
to
the
nur
se o
r cl
inic
al o
ffice
r w
ithin
48
hour
s. o
f rec
eivi
ng
the
text
mes
sage
, oth
erw
ise
the
nurs
e w
ill c
heck
on
them
, gav
e th
em a
n in
cent
ive
to k
eep
on
trac
k of
re
ply
ing,
whi
ch w
as a
qui
ck
and
str
aigh
tfor
war
d p
roce
ss
from
the
ir p
hone
s. T
he
actio
n of
bei
ng t
old
the
y ha
d
to r
esp
ond
with
in a
giv
en
time
initi
ated
the
beh
avio
ur
of t
he c
omm
unic
atio
n b
etw
een
the
pat
ient
and
he
alth
wor
ker.
No
—th
ere
is n
o a
ctua
l ev
iden
ce t
hat
the
wee
kly
rem
ind
er c
ause
d t
he
cont
inue
d o
r im
pro
ved
ad
here
nce
in p
eop
le t
o
take
the
ir t
reat
men
t.
Sha
ring
a p
hone
and
bei
ng
a w
oman
red
uced
the
in
terv
entio
n ef
fect
, but
the
se
fact
ors
are
unex
pla
ined
.
Con
tinue
d
on June 15, 2020 by guest. Protected by copyright.
http://gh.bmj.com
/B
MJ G
lob Health: first published as 10.1136/bm
jgh-2017-000543 on 6 March 2018. D
ownloaded from
Anstey Watkins J, et al. BMJ Glob Health 2018;3:e000543. doi:10.1136/bmjgh-2017-000543 5
BMJ Global Health
Aut
hor,
year
an
d c
oun
try
Res
earc
h d
esig
n, p
arti
cip
ant
sam
ple
, set
ting
, mai
n o
utco
me
mea
sure
and
se
cond
ary
out
com
eIn
terv
enti
on
des
crip
tio
n, in
terv
enti
on
com
po
nent
s an
d s
tud
y d
urat
ion
Inte
rven
tio
n ef
fect
on
pri
mar
y, s
eco
ndar
y an
d p
roce
ss o
utco
mes
and
ove
rvie
w
Aut
hors
’ pro
gra
mm
e th
eory
fo
r th
e in
terv
enti
on
(if s
tate
d) o
r p
rob
able
p
rog
ram
me
theo
ry, b
ased
on
des
crip
tio
n o
f in
terv
enti
on
(incl
udin
g li
kely
m
echa
nism
/int
erve
ntio
n fu
ncti
on)
Did
the
inte
rven
tio
n w
ork
as
inte
nded
or
not?
Bob
row
et
al,
2016
45: S
outh
A
fric
a
Des
ign:
RC
TS
amp
le: P
atie
nts
with
hy
per
tens
ion
(SB
P<
220
mm
H
g an
d a
DB
P<
120
mm
Hg)
ov
er 2
1 ye
ars.
n=
1188
tota
l.S
ettin
g: O
ne u
rban
larg
e p
ublic
hea
lth s
ervi
ce c
linic
.M
ain
outc
ome
mea
sure
: C
hang
e in
sys
tolic
blo
od
pre
ssur
e at
12
mon
ths
from
b
asel
ine
mea
sure
d w
ith a
va
lidat
ed o
scill
omet
ric d
evic
e.S
econ
dar
y ou
tcom
e:
Hea
lth s
tatu
s m
easu
red
us
ing
Eur
oQul
Gro
up
5-D
imen
sion
Sel
f-R
epor
t Q
uest
ionn
aire
(EQ
-5D
), p
rop
ortio
n of
sch
edul
ed
clin
ic a
pp
oint
men
ts a
tten
ded
, re
tent
ion
in c
linic
al c
are,
sa
tisfa
ctio
n w
ith c
linic
se
rvic
es a
nd c
are,
hos
pita
l ad
mis
sion
s, s
elf-
rep
orte
d
adhe
renc
e to
med
icat
ion
(did
th
ey c
olle
ct t
heir
med
icat
ion)
an
d u
nder
stan
din
g of
bas
ic
hyp
erte
nsio
n kn
owle
dge
.
Info
rmat
ion-
only
mes
sage
inte
rven
tion:
A
utom
ated
sys
tem
sen
ds
the
pat
ient
the
tex
t m
essa
ge w
ith p
red
efine
d m
essa
ges
with
he
alth
wor
ker’s
nam
e at
the
end
of m
essa
ge.
Inte
ract
ivity
—w
eekl
y, a
t a
time
pre
defi
ned
b
y th
e p
artic
ipan
t. P
artic
ipan
ts r
ecei
ved
one
m
essa
ge p
er w
eek,
eith
er a
rem
ind
er t
o at
tend
an
up
com
ing
app
oint
men
t (4
8 ho
urs
prio
r to
sch
edul
ed a
pp
oint
men
t) or
a m
essa
ge
sele
cted
-at-
rand
om fr
om t
he m
essa
ge li
bra
ry.
Inte
ract
ive
mes
sage
inte
rven
tion:
Info
rmat
ion-
only
inte
rven
tion
PLU
S p
artic
ipan
ts c
ould
als
o se
lect
mes
sage
s al
loca
ted
to
the
inte
ract
ive
adhe
renc
e su
pp
ort
and
cou
ld a
lso
resp
ond
to
sel
ecte
d m
essa
ges
usin
g fr
ee-t
o-us
er
‘Ple
ase-
Cal
l-M
e’ r
eque
sts.
The
se r
eque
sts
gene
rate
d a
n au
tom
ated
ser
ies
of r
esp
onse
s fr
om t
he t
ext
mes
sage
del
iver
y sy
stem
(not
a
heal
th w
orke
r no
r a
pho
ne c
all r
esp
onse
) of
ferin
g tr
ial p
artic
ipan
ts o
ptio
ns t
o ca
ncel
or
chan
ge a
n ap
poi
ntm
ent
or c
hang
e th
e tim
ing
and
lang
uage
of t
he t
ext
mes
sage
s. P
atie
nts
also
rec
eive
d a
tex
t m
essa
ge w
ere
sent
to
eith
er t
hank
par
ticip
ants
for
atte
ndin
g th
eir
app
oint
men
t or
ale
rt p
artic
ipan
ts a
bou
t a
mis
sed
ap
poi
ntm
ent
48 h
ours
. pos
tdat
e.Th
ere
is n
o p
erso
nal c
onta
ct b
etw
een
nurs
e an
d p
artic
ipan
t.In
terv
entio
n co
mp
onen
ts:
Rem
ind
ers
Targ
eted
act
ions
Stu
dy
dur
atio
n: 1
2 m
onth
s (b
asel
ine,
follo
w-u
p
at 6
mon
ths
and
12
mon
ths)
.
Imp
act
outc
ome:
The
mea
n (9
5% C
I, p
val
ue)
adju
sted
dif f
eren
ce in
cha
nge
for
the
info
rmat
ion-
only
mes
sage
gro
up c
omp
ared
w
ith u
sual
car
e w
as s
ysto
lic b
lood
pre
ssur
e −
2.2
mm
Hg
(−4.
4 to
−0.
04, p
=0.
046)
and
for
the
inte
ract
ive
mes
sage
gr o
up c
omp
ared
with
us
ual c
are
−1.
6 m
m H
g (−
3.7
to 0
.6, p
=0.
16).
Sec
ond
ary
outc
omes
: Out
of t
he 8
6%
of p
atie
nts,
the
tria
l had
ad
her e
nce
dat
a av
aila
ble
, 63%
from
the
info
rmat
ion
only
m
essa
ge in
terv
entio
n gr
oup
, 60%
for
the
inte
ract
ive
mes
sage
inte
rven
tion
grou
p a
nd
49%
from
the
usu
al c
are
grou
p, h
ad 8
0%
of p
rop
ortio
n of
day
s co
vere
d fo
r b
lood
p
ress
ure
low
erin
g m
edic
atio
n fo
r a
12-m
onth
p
erio
d. E
Q-5
D s
core
s, a
tten
dan
ce a
t cl
inic
ap
poi
ntm
ents
, ret
entio
n in
clin
ical
car
e,
trea
tmen
t an
d c
linic
sat
isfa
ctio
n, h
yper
tens
ion
know
led
ge, s
elf-
rep
orte
d a
dhe
renc
e, h
osp
ital
adm
issi
ons
and
diff
eren
ces
in m
edic
atio
n ch
ange
s d
id n
ot d
iffer
bet
wee
n gr
oup
s.S
ubgr
oup
ana
lysi
s: T
here
was
no
stat
istic
ally
si
gnifi
cant
het
erog
enei
ty in
the
tre
atm
ent
effe
cts
and
the
re w
as a
n in
dic
atio
n th
at a
ctiv
e in
terv
entio
ns w
ere
mor
e ef
fect
ive
amon
g ol
der
p
atie
nts
(55
year
s), p
atie
nts
in b
ette
r co
ntro
l at
bas
elin
e (<
140
mm
Hg)
and
am
ong
thos
e w
ith
a sh
orte
r d
urat
ion
of h
yper
tens
ion
(<10
yea
rs).
By
r ece
ivin
g a
beh
avio
ural
inte
rven
tion
del
iver
ed v
ia t
ext
mes
sage
as
sup
por
t, t
his
coul
d im
pro
ve c
olle
ctio
n of
med
icin
es a
nd
may
hav
e a
smal
l im
pac
t on
blo
od p
ress
ure
as c
omp
ared
with
usu
al c
are
in a
gen
eral
ou
tpat
ient
pop
ulat
ion
of a
dul
ts w
ith h
igh
blo
od p
ress
ure.
Get
ting
info
rmat
ion
to
peo
ple
at
a tim
e th
at is
rel
evan
t to
the
m
and
pro
mp
ts t
o ta
ke m
edic
atio
n sh
ould
he
lp t
o en
cour
age
the
pat
ient
bec
ause
the
y ha
ve c
hose
n w
hen
to r
ecei
ve t
he m
essa
ge
and
thi
s tim
e is
mos
t ap
pro
pria
te t
o th
eir
dai
ly r
outin
e. R
emin
der
mes
sage
s w
ith
info
rmat
ion
abou
t fo
rthc
omin
g or
mis
sed
cl
inic
ap
poi
ntm
ents
rem
ind
the
pat
ient
ab
out
whe
re t
hey
need
to
be
and
whe
n.R
emin
der
s ab
out
who
the
y ca
n co
ntac
t if
they
are
wor
ried
or
conc
erne
d a
bou
t an
y si
de
effe
cts
of t
he m
edic
atio
n gi
ves
the
pat
ient
incr
ease
d k
now
led
ge a
nd c
hann
els
of s
upp
ort.
By
rece
ivin
g a
‘Hap
py
Birt
hday
’ m
essa
ge t
his
mak
es t
he p
atie
nt fe
el c
ared
fo
r b
y th
e he
alth
ser
vice
, on
thei
r d
ay o
f b
irth.
Stu
dy
des
ign
issu
es: T
he in
terv
entio
n in
clud
es b
ehav
iour
cha
nge
clus
ter
tech
niq
ues
such
as
rep
etiti
on a
nd
sub
stitu
tion,
goa
ls a
nd p
lann
ing
and
soc
ial
sup
por
t w
hich
are
bas
ed o
n M
ichi
e's
Beh
avio
ur C
hang
e W
heel
.
No
—th
e p
rim
ary
out
com
e d
id n
ot
imp
rove
.B
lood
pre
ssur
e le
vels
in
bot
h in
terv
entio
n gr
oup
s d
id n
ot d
ecre
ase
dur
ing
the
tria
l per
iod
sug
gest
ing
the
text
mes
sage
(inf
orm
atio
n an
d in
tera
ctiv
e) d
id n
ot h
elp
to
red
uce
blo
od p
ress
ure
cont
rol i
n ad
ults
with
d
iagn
osed
hig
h B
P a
nd o
n tr
eatm
ent
(par
tly b
ecau
se
man
y of
the
m w
ere
stab
le a
t b
asel
ine)
.Ye
s—th
e se
cond
ary
out
com
es d
id im
pro
ve.
Pro
por
tion
of d
ays
of
med
icat
ion
cove
red
(a
dhe
renc
e to
med
icat
ion)
w
as h
ighe
r in
bot
h in
terv
entio
n gr
oup
s su
gges
ting
that
the
y w
ere
rem
emb
erin
g to
col
lect
th
eir
med
icat
ion
bec
ause
of
the
rem
ind
ers:
the
tw
o in
terv
entio
n gr
oup
s w
ere
colle
ctin
g th
e m
edic
atio
n m
ore
regu
larly
ove
r th
e co
ntro
l gro
ups.
Tab
le 2
C
ontin
ued
Con
tinue
d
on June 15, 2020 by guest. Protected by copyright.
http://gh.bmj.com
/B
MJ G
lob Health: first published as 10.1136/bm
jgh-2017-000543 on 6 March 2018. D
ownloaded from
6 Anstey Watkins J, et al. BMJ Glob Health 2018;3:e000543. doi:10.1136/bmjgh-2017-000543
BMJ Global Health
Aut
hor,
year
an
d c
oun
try
Res
earc
h d
esig
n, p
arti
cip
ant
sam
ple
, set
ting
, mai
n o
utco
me
mea
sure
and
se
cond
ary
out
com
eIn
terv
enti
on
des
crip
tio
n, in
terv
enti
on
com
po
nent
s an
d s
tud
y d
urat
ion
Inte
rven
tio
n ef
fect
on
pri
mar
y, s
eco
ndar
y an
d p
roce
ss o
utco
mes
and
ove
rvie
w
Aut
hors
’ pro
gra
mm
e th
eory
fo
r th
e in
terv
enti
on
(if s
tate
d) o
r p
rob
able
p
rog
ram
me
theo
ry, b
ased
on
des
crip
tio
n o
f in
terv
enti
on
(incl
udin
g li
kely
m
echa
nism
/int
erve
ntio
n fu
ncti
on)
Did
the
inte
rven
tio
n w
ork
as
inte
nded
or
not?
Pie
tte
et a
l, 20
1243
: H
ond
uras
and
M
exic
o
Des
ign:
RC
TS
amp
le: P
atie
nts
with
hy
per
tens
ion
(‡14
0 m
m H
g if
non-
dia
bet
ic) o
r hy
per
tens
ion
and
dia
bet
es*
(‡13
0 m
m H
g if
dia
bet
ic) 1
8–80
yea
rs o
f ag
e. In
terv
entio
n gr
oup
(tex
t m
essa
ge) a
nd p
rep
lann
ed
sub
grou
p a
naly
sis:
pat
ient
s w
ith lo
w li
tera
cy o
r hi
gh B
P
man
agem
ent
info
rmat
ion
need
s (n
=89
). C
ontr
ol g
roup
(s
tand
ard
car
e) (n
=92
).*B
lood
glu
cose
was
not
bei
ng
mea
sure
d in
thi
s st
udy.
Set
ting:
Fou
r p
rivat
e an
d t
wo
pub
lic c
linic
s in
Hon
dur
as a
nd
two
clin
ics
in M
exic
o.M
ain
outc
ome
mea
sure
: C
hang
e in
sys
tolic
blo
od
pre
ssur
e (S
BP
s ‡1
60 m
m H
g an
d<
180
mm
Hg)
.S
econ
dar
y ou
tcom
es:
Pat
ient
s’ p
erce
ived
gen
eral
he
alth
sta
tus,
dep
ress
ive
sym
pto
ms
(usi
ng a
va
lidat
ed S
pan
ish
10-i
tem
C
entr
e fo
r E
pid
emio
logi
cal
Stu
die
s-D
epre
ssio
n S
cale
), m
edic
atio
n-re
late
d p
rob
lem
s (a
dhe
renc
e m
easu
red
usi
ng
the
Mor
isky
Sca
le) a
nd
satis
fact
ion
with
car
e re
late
d
to h
yper
tens
ion.
Aut
omat
ed t
elep
hone
mon
itorin
g an
d
beh
avio
ur-c
hang
e ca
lls p
lus
hom
e B
P
mon
itorin
g am
ong
hyp
erte
nsiv
e p
atie
nts.
The
ca
lls w
ere
aim
ed a
t ga
ther
ing
info
rmat
ion
abou
t th
e p
atie
nt’s
BP,
BP
sel
f-m
onito
ring,
med
icat
ion
adhe
renc
e an
d d
iet
and
to
pro
vid
e ta
ilore
d
advi
ce b
ased
on
the
pat
ient
’s r
esp
onse
s.P
atie
nts’
with
hyp
erte
nsio
n re
ceiv
ed a
hom
e b
lood
pre
ssur
e m
onito
r an
d w
ere
give
n tr
aini
ng. T
he in
terv
entio
n fo
cuse
d m
ainl
y on
p
rovi
din
g in
form
atio
n an
d s
elf-
man
agem
ent
educ
atio
n to
pat
ient
s vi
a in
tera
ctiv
e vo
ice
calls
or
auto
mat
ed c
alls
—to
che
ck t
heir
BP
an
d w
ere
aske
d a
bou
t re
cent
sys
tolic
val
ues
abov
e an
d b
elow
the
nor
mal
ran
ge, m
edic
atio
n ad
here
nce
and
inta
ke o
f sal
ty fo
ods.
Bas
ed o
n w
hat
they
sai
d, t
he p
atie
nts
wer
e th
en o
ffere
d
add
ition
al s
elf-
care
info
rmat
ion
dur
ing
the
call
and
pro
mp
ts t
o se
ek m
edic
al a
tten
tion
or
med
icat
ion
refil
ls t
o ad
dre
ss u
nacc
epta
bly
hig
h or
low
BP.
Str
uctu
red
em
ail a
lert
s fo
r he
alth
w
orke
rs w
ere
gene
rate
d a
utom
atic
ally
whe
n p
atie
nts
rep
orte
d t
hat
at le
ast
half
the
time
in
the
prio
r w
eek
they
had
an
SB
P ‡
140
mm
Hg
(pat
ient
s w
ith n
on-d
iab
etes
), ‡
130
mm
Hg
(dia
bet
ic p
atie
nts)
or
100
mm
Hg
(all
pat
ient
s)
or if
the
pat
ient
rep
orte
d r
arel
y or
nev
er t
akin
g hi
s or
her
BP
med
icat
ion
or le
ss t
han
a 2
wee
k su
pp
ly. P
atie
nts
had
the
op
tion
of e
nrol
ling
with
a fa
mily
mem
ber
or
frie
nd, w
ho r
ecei
ved
a
brie
f aut
omat
ed t
elep
hone
up
dat
e re
gard
ing
the
pat
ient
’s s
elf-
rep
orte
d h
ealth
sta
tus
each
w
eek,
incl
udin
g in
form
atio
n ab
out
the
pat
ient
’s
hyp
erte
nsio
n se
lf-ca
re a
nd h
ow t
hat
care
give
r co
uld
hel
p t
he p
atie
nt s
elf-
man
age
mor
e ef
fect
ivel
y.In
terv
entio
n co
mp
onen
ts:
Ob
serv
atio
n
R
emin
der
s
M
otiv
atin
g ed
ucat
ion/
advi
ce in
form
atio
nS
tud
y d
urat
ion:
6 w
eeks
(with
op
tion
of a
3
mon
th e
xten
sion
).
Imp
act
outc
ome:
Inte
rven
tion
pat
ient
s’ S
BP
s d
ecre
ased
4.2
mm
Hg
rela
tive
to c
ontr
ols
(95%
CI 9
.1 t
o 0.
7; p
=0.
09).
In t
he s
ubgr
oup
w
ith h
igh
info
rmat
ion
need
s, in
terv
entio
n p
atie
nts’
ave
rage
SB
Ps
dec
reas
ed 8
.8 m
m H
g (–
14.2
to
–3.4
, p=
0.00
2). 5
7% o
f int
erve
ntio
n p
atie
nts
had
con
trol
led
BP
at
follo
w-u
p
com
par
ed w
ith 3
8% o
f the
com
par
ison
gro
up
(p=
0.00
6).
Pro
cess
out
com
e: M
ore
than
88%
of p
atie
nts
rep
orte
d t
hat
the
auto
mat
ed c
allin
g sy
stem
w
as e
asy
to le
arn
and
use
, and
93%
rep
orte
d
that
the
aut
omat
ed c
alls
incl
uded
use
ful
info
rmat
ion
for
man
agin
g th
eir
hyp
erte
nsio
n.
Ove
rall,
94%
of i
nter
vent
ion
pat
ient
s re
por
ted
b
eing
ver
y sa
tisfie
d w
ith t
he in
terv
entio
n an
d 7
6% r
epor
ted
tha
t th
e p
rogr
amm
e w
as
exce
llent
.S
econ
dar
y ou
tcom
es: A
t fo
llow
-up
, pat
ient
s ha
d lo
wer
dep
ress
ive
scor
es a
nd fe
wer
m
edic
atio
n-re
late
d p
rob
lem
s fo
r ex
amp
le,
wor
ry a
bou
t th
e ef
fect
s of
the
ir m
edic
atio
n.
Ove
rall
heal
th w
as r
epor
ted
and
the
re w
as
grea
ter
satis
fact
ion
with
car
e-re
late
d t
o hy
per
tens
ion.
Sub
grou
p a
naly
sis:
Pat
ient
s w
ith lo
w
liter
acy
or h
igh
BP
man
agem
ent
info
rmat
ion
need
s: t
hose
with
pro
ble
ms
lear
ning
ab
out
thei
r he
alth
pro
ble
ms
bec
ause
of d
ifficu
lty
und
erst
and
ing
writ
ten
info
rmat
ion,
had
nev
er
bee
n to
ld t
hey
had
hyp
erte
nsio
n or
had
not
sp
oken
with
a c
linic
ian
abou
t th
eir
BP
in m
ore
than
6 m
onth
s or
wer
e co
nfus
ed a
bou
t th
eir
med
icat
ion
regi
men
. The
y ha
d a
n av
erag
e of
8.8
mm
Hg
red
uctio
n in
SB
P r
elat
ive
to
cont
rols
(95%
CI
14.2
to
–3.4
; p=
0.00
2).
Out
com
e ov
ervi
ew: I
n th
e ov
eral
l sam
ple
, th
ere
was
a n
on-s
tatis
tical
ly s
igni
fican
t (p
=0.
09) 4
.2 m
m H
g re
lativ
e d
ecre
ase
in
SB
P a
mon
g in
terv
entio
n p
atie
nts.
In t
he
sub
grou
p o
f pat
ient
s w
ith lo
w li
tera
cy o
r hi
gh
info
rmat
ion
need
s, 8
.8 m
m H
g re
duc
tion
in
aver
age
SB
P w
as o
bse
rved
with
a s
igni
fican
tly
grea
ter
pro
por
tion
of in
terv
entio
n th
an c
ontr
ol
pat
ient
s ha
ving
BP
s in
the
acc
epta
ble
ran
ge.
Inte
rven
tion
pat
ient
s at
follo
w-u
p h
ad S
BP
s th
at w
ere
4.2
mm
Hg
low
er o
n av
erag
e th
an
cont
rol p
atie
nts
(95%
CI
9.1
to 0
.7; p
=0.
09).
By
rece
ivin
g au
tom
ated
sel
f-m
anag
emen
t ca
lls, p
lus
hom
e b
lood
pre
ssur
e m
onito
ring
kit,
thi
s ca
n im
pro
ve o
utco
mes
for
hyp
erte
nsiv
e p
atie
nts
as r
emin
der
s to
che
ck
blo
od p
ress
ure
read
ings
sev
eral
tim
es p
er
wee
k th
is a
cts
a nu
dge
to
actio
n.Th
e m
echa
nism
s of
act
ion
incl
uded
:(1
) Dur
ing
the
calls
pat
ient
s w
ere
rem
ind
ed
to c
heck
the
ir B
P r
egul
arly
and
wer
e as
ked
ab
out
rece
nt s
ysto
lic v
alue
s ab
ove
and
bel
ow t
he n
orm
al r
ange
, med
icat
ion
adhe
renc
e, a
nd in
take
of s
alty
food
s. T
his
regu
lar
chec
king
-in
mea
nt t
he p
atie
nt h
ad
to k
eep
on
top
of t
heir
man
agem
ent
and
re
gula
rly v
erb
ally
dis
cuss
the
ir ch
roni
c d
isea
se.
(2) T
he h
ealth
wor
kers
wer
e al
erte
d v
ia e
mai
l if
a p
atie
nt’s
blo
od p
ress
ure
chan
ged
or
if m
edic
atio
n w
as n
ot t
aken
. The
refo
re, t
he
pat
ient
s kn
ew t
hat
thei
r re
sults
wer
e go
ing
to b
e re
por
ted
if t
hey
faile
d t
o ta
ke t
heir
BP
m
edic
atio
n. T
his
crea
ted
an
ince
ntiv
e to
ad
here
.(3
) The
sup
por
t of
a fa
mily
mem
ber
or
frie
nd m
eant
tha
t th
e p
atie
nt’s
sel
f-re
por
ted
he
alth
sta
tus
and
sel
f-ca
re w
ere
give
n to
the
tr
eatm
ent
sup
por
ter.
This
pro
cess
mea
nt t
he
pat
ient
was
acc
ount
able
to
som
eone
els
e.
Ano
ther
per
son
was
par
t of
the
ir ch
roni
c d
isea
se m
anag
emen
t.Th
e in
form
atio
n-b
ased
inte
rven
tion
had
gr
eate
r im
pac
t on
pat
ient
s w
ho r
epor
ted
a
grea
ter
need
for
hyp
erte
nsio
n-re
late
d
know
led
ge a
nd e
duc
atio
n b
ecau
se t
hey
had
lo
w li
tera
cy o
r hi
gh in
form
atio
n ne
eds
and
va
lued
ad
diti
onal
sup
por
tive
info
rmat
ion.
Yes—
blo
od
pre
ssur
e d
ecre
ased
. Cal
ls a
re
effe
ctiv
e ov
er a
sho
rt fo
llow
-up
per
iod
.Ye
s—p
arti
cula
rly
for
the
sub
gro
ups
wit
h lo
w
liter
acy
and
hig
h B
P. T
he
inte
rven
tion
focu
sed
on
pro
vid
ing
info
rmat
ion
and
se
lf-m
anag
emen
t ed
ucat
ion
to p
atie
nts.
Tab
le 2
C
ontin
ued
Con
tinue
d
on June 15, 2020 by guest. Protected by copyright.
http://gh.bmj.com
/B
MJ G
lob Health: first published as 10.1136/bm
jgh-2017-000543 on 6 March 2018. D
ownloaded from
Anstey Watkins J, et al. BMJ Glob Health 2018;3:e000543. doi:10.1136/bmjgh-2017-000543 7
BMJ Global Health
Aut
hor,
year
an
d c
oun
try
Res
earc
h d
esig
n, p
arti
cip
ant
sam
ple
, set
ting
, mai
n o
utco
me
mea
sure
and
se
cond
ary
out
com
eIn
terv
enti
on
des
crip
tio
n, in
terv
enti
on
com
po
nent
s an
d s
tud
y d
urat
ion
Inte
rven
tio
n ef
fect
on
pri
mar
y, s
eco
ndar
y an
d p
roce
ss o
utco
mes
and
ove
rvie
w
Aut
hors
’ pro
gra
mm
e th
eory
fo
r th
e in
terv
enti
on
(if s
tate
d) o
r p
rob
able
p
rog
ram
me
theo
ry, b
ased
on
des
crip
tio
n o
f in
terv
enti
on
(incl
udin
g li
kely
m
echa
nism
/int
erve
ntio
n fu
ncti
on)
Did
the
inte
rven
tio
n w
ork
as
inte
nded
or
not?
Kam
al e
t al
, 20
1542
: P
akis
tan
Des
ign:
RC
TS
amp
le: P
atie
nts
with
str
oke
18
yea
rs a
nd o
ver.
>1
mon
th
sinc
e la
st e
pis
ode
of s
trok
e.
Use
of a
t le
ast
two
dru
gs
such
as
(but
not
lim
ited
to)
an
tipla
tele
ts, s
tatin
s an
d
antih
yper
tens
ives
to
cont
rol
risk
fact
ors
of s
trok
e (n
=83
). C
ontr
ol g
roup
(n=
79).
Set
ting:
One
urb
an h
osp
ital’s
ne
urol
ogy
and
str
oke
unit.
Mai
n ou
tcom
e m
easu
re:
Cha
nge
in s
elf-
rep
orte
d s
trok
e m
edic
atio
n ad
here
nce
afte
r 2
mon
ths
of r
ecei
ving
the
tex
t m
essa
ge (u
sing
the
Mor
isky
S
cale
).S
econ
dar
y ou
tcom
e: B
lood
p
ress
ure
was
mea
sure
d
usin
g th
e M
ind
ray
Dat
asco
pe
Eq
uato
r to
det
ect
chan
ge in
sy
stol
ic a
nd D
BP.
Aut
omat
ed t
ext
mes
sage
rem
ind
ers
cust
omis
ed t
o ea
ch p
atie
nt’s
ind
ivid
ual
pre
scrip
tion.
Pat
ient
s w
ere
req
uire
d t
o re
spon
d
to t
he t
ext
mes
sage
, sta
ting
if th
ey h
ad t
aken
th
eir
med
icin
es b
y re
ply
ing
‘Yes
’ or
‘No’
. Als
o,
cust
omis
ed t
wic
e-w
eekl
y he
alth
info
rmat
ion
text
mes
sage
s w
ere
sent
acc
ord
ing
to m
edic
al
and
dru
g p
rofil
e of
eve
ry p
atie
nt. T
he t
imin
g of
the
mes
sage
was
dec
ided
acc
ord
ing
to t
he
pre
scrip
tion
so t
hat
heal
th m
essa
ges
did
not
co
llid
e w
ith r
emin
der
mes
sage
s.In
terv
entio
n co
mp
onen
ts:
Rem
ind
ers
Mot
ivat
ing
educ
atio
n/ad
vice
info
rmat
ion
Pra
ise
and
enc
oura
gem
ent
Stu
dy
dur
atio
n: B
asel
ine,
2 m
onth
s fo
llow
-up
.
Imp
act
outc
ome:
Mea
n d
iffer
ence
in
adhe
renc
e sc
ore
bet
wee
n th
e in
terv
entio
n gr
oup
and
the
con
trol
gro
up (u
sual
car
e) w
as
0.54
(95
% C
I 0.2
2 to
0.8
5; p
≤0.0
1).
Pro
cess
out
com
e: P
atie
nt s
atis
fact
ion
and
ac
cep
tab
ility
usi
ng t
ext
mes
sage
s to
imp
rove
cl
inic
al o
utco
me,
was
96%
aft
er t
he 2
-mon
th
per
iod
.S
econ
dar
y ou
tcom
e: R
educ
tion
in D
BP.
No
maj
or e
ffect
was
ob
serv
ed o
n sy
stol
ic b
lood
p
ress
ure
due
to
reas
onab
ly li
mite
d e
xpos
ure
time
to t
he in
terv
entio
n.O
verv
iew
out
com
e: A
sig
nific
ant
affe
ct w
as
foun
d in
the
inte
rven
tion
grou
p's
ad
here
nce
to s
trok
e m
edic
atio
n. It
was
foun
d t
hat
bei
ng
retir
ed a
nd/o
r un
emp
loye
d, b
eing
ed
ucat
ed
and
with
a h
ighe
r d
osin
g fr
eque
ncy,
wer
e p
ositi
vely
rel
ated
to
the
leve
l of m
edic
atio
n ad
here
nce
foun
d.
By
cust
omis
ing
the
mes
sage
s, t
he p
atie
nts
are
likel
y to
be
mor
e sa
tisfie
d t
o ac
t as
th
ey w
ere
aske
d t
o re
ply
with
a v
ery
sim
ple
re
spon
se. T
ext
mes
sage
s w
ere
cust
omis
ed
for
each
pat
ient
dep
end
ing
on t
heir
med
icat
ion
pre
scrip
tion
(com
par
ed w
ith
just
rec
eivi
ng s
imp
le k
now
led
ge t
rans
fer
mes
sage
s). T
he p
atie
nts
wer
e b
eing
tre
ated
th
eref
ore
as s
ingl
e ca
ses.
Als
o, t
he t
imin
g of
the
mes
sage
s se
nt o
ut a
ccor
din
g to
the
p
atie
nt's
dos
ing
sche
dul
e w
as im
por
tant
in
incr
easi
ng a
dhe
renc
e. T
his
timin
g ta
rget
ed
an a
ctio
n b
ecau
se t
hey
wer
e re
ceiv
ed a
t a
time
app
rop
riate
to
the
per
son.
The
se
mes
sage
s ta
rget
ed b
oth
inte
ntio
nal
non-
adhe
renc
e an
d n
on-i
nten
tiona
l non
-ad
here
nce
by
pro
vid
ing
know
led
ge a
nd
bel
ief c
hang
e m
essa
ges
and
oth
er c
uing
, nu
dgi
ng a
nd r
emin
der
beh
avio
urs
to t
ake
med
icat
ions
. Rem
ind
ers
abou
t b
ehav
iour
w
ere
likel
y to
ent
ice
the
pat
ient
to
take
the
ir m
edic
atio
n al
ong
with
giv
ing
pra
ise
and
en
cour
agem
ent.
Stu
dy
des
ign
issu
es:
Con
tent
and
lang
uage
of i
nter
vent
ion
mes
sage
s ar
e d
esig
ned
on
The
Hea
lth
Bel
ief M
odel
and
Soc
ial C
ogni
tive
Theo
ry.
Yes—
in t
he 2
mo
nths
st
udie
d, a
dhe
r enc
e im
pro
ved
. The
re w
as a
lso
a
slig
ht r
educ
tio
n in
DB
P.
AR
T, a
ntire
trov
iral t
hera
py;
BP,
blo
od p
ress
ure;
DB
P, d
iast
olic
blo
od p
ress
ure;
RC
T, r
and
omis
ed c
ontr
olle
d t
rial;
RR
, Rel
ativ
e R
isk;
SB
P, s
ysto
lic b
lood
pre
ssur
e.
Tab
le 2
C
ontin
ued
on June 15, 2020 by guest. Protected by copyright.
http://gh.bmj.com
/B
MJ G
lob Health: first published as 10.1136/bm
jgh-2017-000543 on 6 March 2018. D
ownloaded from
8 Anstey Watkins J, et al. BMJ Glob Health 2018;3:e000543. doi:10.1136/bmjgh-2017-000543
BMJ Global Health
Tab
le 3
In
terv
entio
n co
mp
onen
ts d
escr
ibed
in r
elat
ion
to e
stab
lishe
d t
heor
y in
all
four
incl
uded
em
piri
cal s
tud
ies
on m
Hea
lth in
Ken
ya, S
outh
Afr
ica,
Hon
dur
as a
nd
Mex
ico
and
Pak
ista
n
Co
mp
one
nts
of
the
inte
rven
tio
nE
xam
ple
to
exp
lain
the
inte
rven
tio
n co
mp
one
nts
fro
m
stud
y in
terv
enti
ons
Est
ablis
hed
the
ory
tha
t re
late
s to
one
of
mo
re o
f th
e in
terv
enti
on
com
po
nent
s
Typ
es o
f in
terv
enti
on
com
po
nent
s us
ed
by
each
stu
dy
auth
or
in t
he d
esig
n o
f th
eir
mH
ealt
h in
terv
enti
on
Form
of c
omm
unic
atio
n
1.
Two-
way
com
mun
icat
ion
2.
Whe
ther
the
com
mun
icat
ion
and
res
pon
se
from
the
hea
lth s
yste
m is
aut
omat
ed o
r no
t (is
the
re a
rea
l per
son
enga
ging
with
th
e p
atie
nt)
3.
Imp
orta
nce
of p
erso
nal c
onta
ct fo
r m
otiv
atio
n an
d s
upp
ort
This
tw
o-w
ay c
omm
unic
atio
n w
hich
is a
utom
ated
or
not
may
ena
ble
the
pro
visi
on o
f sup
por
t. S
upp
ort
is t
he c
onte
nt
whi
ch fl
ows
thro
ugh
the
chan
nel o
f com
mun
icat
ion
and
is
sep
arat
e fr
om t
ype
of c
omm
unic
atio
n.P
atie
nt r
ecei
ves
tailo
red
dig
ital c
ouns
ellin
g an
d fe
edb
ack
afte
r su
bm
ittin
g cl
inic
al m
easu
rem
ents
. Thi
s en
able
s th
e es
tab
lishm
ent
of a
rel
atio
nshi
p w
ith h
ealth
wor
ker
and
in
crea
ses
the
acce
ss t
o th
e su
pp
ort.
The
text
mes
sage
or
voic
e ca
ll m
otiv
ates
the
pat
ient
by
info
rmin
g th
em t
o ea
t m
ore
vege
tab
les
or r
educ
e sa
lt in
take
. The
des
ire t
o ca
rry
out
the
beh
avio
ur c
hang
e or
ad
here
nce
to t
reat
men
t m
otiv
ated
by
thei
r b
elie
fs,
exp
ecta
tions
and
feel
ings
.
1.
Hea
lth C
omm
unic
atio
n Th
eory
/Acc
ess
Fram
ewor
k62:
whe
reb
y us
ing
mH
ealth
enh
ance
s ac
cess
to
sup
por
t fr
om a
hea
lth w
orke
r.2.
In
form
atio
n-M
otiv
atio
n-S
trat
egy
Mod
el63
: whe
reb
y ad
here
nce
to t
reat
men
t re
qui
res
pat
ient
s to
hav
e (1
)In
form
atio
n, (2
) Mot
ivat
ion
and
(3) S
trat
egie
s to
ad
here
to
tre
atm
ent
givi
ng t
he p
atie
nt t
he in
cent
ive
to b
ehav
e ap
pro
pria
tely
and
acc
ord
ing
to t
heir
regi
men
.3.
Th
eory
of I
nter
per
sona
l Rel
atio
nshi
ps64
: whe
reb
y in
terp
erso
nal i
nter
actio
ns b
etw
een
the
pat
ient
and
th
e he
alth
wor
ker
can
help
with
tas
k en
gage
men
t.
This
the
ory
seek
s to
und
erst
and
whi
ch c
hara
cter
istic
s st
reng
then
the
bon
ds
bet
wee
n p
eop
le, e
ncou
rage
em
pat
hy a
nd t
rust
and
cre
ate
a se
nse
of w
ell-
bei
ng
afte
r a
clos
e in
tera
ctio
n to
influ
ence
hum
an m
otiv
atio
n.
The
stro
nger
the
mot
ivat
ion,
the
str
onge
r th
e re
latio
nshi
p b
ecom
es, b
ecau
se e
mp
athy
and
car
e ca
n d
eriv
e fr
om t
he in
terp
erso
nal p
roce
sses
.
Two-
way
com
mun
icat
ion:
Kam
al, L
este
r, P
iett
e,
Bob
row
Aut
omat
ed: B
obro
w, K
amal
Rea
l per
son:
Les
ter,
Pie
tte
Act
iviti
es t
hat
com
mun
icat
ion
chan
nel i
s us
ed
for:
1.
Ob
serv
atio
n2.
P
rovi
din
g in
form
atio
n
Ob
serv
atio
n: P
atie
nt is
ob
serv
ed in
dire
ctly
and
mus
t re
gula
rly s
end
clin
ical
rea
din
gs. I
f the
pat
ient
is b
eing
ob
serv
ed t
his
acts
as
per
suas
ion
to m
aint
ain
mon
itorin
g as
th
ey a
re n
ow a
ccou
ntab
le t
o so
meo
ne.
Info
rmat
ion:
The
tex
t m
essa
ge o
r vo
ice
call
cont
ains
d
isea
se-s
pec
ific
info
rmat
ion
aim
ed a
t in
crea
sing
kn
owle
dge
. Inc
reas
ing
pat
ient
kno
wle
dge
thr
ough
mob
ile
info
rmat
ion
rem
ind
er o
r le
arni
ng m
essa
ges
thro
ugh
reco
mm
end
atio
ns.
1.
Com
mun
icat
ion
for
Per
suas
ion65
: whe
reb
y th
e p
atie
nt
can
be
per
suad
ed b
y a
role
mod
el w
him
the
y fe
el
acco
unta
ble
to.
2.
Hea
lth B
elie
f Mod
el66
: whe
reb
y a
pat
ient
nee
ds
suffi
cien
t m
otiv
atio
n to
mak
e th
e he
alth
issu
e re
leva
nt,
the
bel
ief i
t is
a s
erio
us p
rob
lem
and
to
follo
w a
re
com
men
dat
ion
to r
educ
e th
at t
hrea
t.
Ob
serv
atio
n: L
este
r, P
iett
e
Diff
eren
t ty
pes
of i
nfor
mat
ion
1.
Rem
ind
ers
2.
Ed
ucat
ion/
advi
ce3.
Ta
rget
ed a
ctio
ns4.
P
rais
e an
d e
ncou
rage
men
t
Rem
ind
ers:
Rem
ind
er m
essa
ges
to a
tten
d c
linic
ap
poi
ntm
ents
to
imp
rove
ret
entio
n to
car
e.R
emin
der
mes
sage
s to
tak
e m
edic
atio
n to
incr
ease
ad
here
nce.
Rem
ind
er c
an a
ct a
s a
cue
to a
ctio
n to
stim
ulat
e th
e p
atie
nt in
to a
ctio
n.R
emin
der
s ca
n he
lp t
o re
info
rce
the
pat
ient
s’ t
houg
hts
and
feel
ings
tha
t le
ad t
o th
eir
dec
isio
ns (a
ttitu
de
tow
ard
s b
ehav
iour
).E
duc
atio
n/ad
vice
: If a
pat
ient
list
ens
to e
duc
atio
nal
advi
ce g
iven
by
a he
alth
pro
fess
iona
l via
tex
t or
voi
ce, i
t is
pos
sib
le t
hat
if th
e p
atie
nt is
cap
able
of u
nder
stan
din
g w
hat
they
hav
e re
ad o
r he
ard
, the
y ca
n th
en fo
llow
the
ad
vice
to
chan
ge t
heir
beh
avio
ur.
Targ
eted
act
ions
: Dig
ital m
obile
dec
isio
n-su
pp
ort
is li
kely
to
aid
the
acc
omp
lishm
ent
of g
oals
. Set
ting
a go
al s
uch
as
taki
ng m
edic
atio
n or
ob
serv
ing
a m
ood
sta
te.
Pra
ise
and
enc
oura
gem
ent:
The
tex
t m
essa
ge o
r vo
ice
call
is in
tend
ed t
o p
rom
ote
good
hea
lthily
beh
avio
ur a
nd
mak
ing
the
pat
ient
feel
op
timis
tic 'Y
ou a
re d
oing
wel
l'. If
a
pat
ient
rec
eive
s a
call
or m
essa
ge a
bou
t ad
optin
g a
heal
thy
lifes
tyle
, thi
s p
rom
otes
a b
ehav
iour
al c
hang
e.
1.
Theo
ry o
f Pla
nned
Beh
avio
ur67
: whe
reb
y th
e p
atie
nt is
m
ade
awar
e of
the
ir th
ough
ts a
nd fe
elin
gs t
hat
lead
to
dec
isio
ns.
2.
Cog
nitiv
e Lo
ad T
heor
y68: w
here
by
the
dem
and
s of
a
cert
ain
task
on
the
pat
ient
acc
ount
s fo
r th
eir
own
bel
iefs
, exp
ecta
tions
and
goa
ls o
n th
eir
own
load
p
erce
ptio
ns.
3.
Goa
l Set
ting
Theo
ry69
: whe
reb
y th
e d
ifficu
lty o
f the
tas
k is
med
iate
d b
ecau
se t
he p
atie
nt is
wor
king
tow
ard
s a
goal
. Fee
db
ack
is a
lso
usef
ul.
4.
Sel
f-R
egul
atio
n Th
eory
70: w
here
by
attit
ude
chan
ge c
an
resu
lt w
hen
the
pat
ient
rea
lises
and
ob
serv
es t
heir
own
beh
avio
urs,
whi
ch c
an in
clud
e m
onito
ring
thei
r m
ood
.
Rem
ind
ers:
Kam
al, L
este
r, P
iett
e, B
obro
wE
duc
atio
n/ad
vice
: Pie
tte,
Kam
alP
rovi
sion
of s
upp
ort
com
mun
icat
ion:
Les
ter
Targ
eted
act
ions
: Bob
row
Pra
ise
and
enc
oura
gem
ent:
Kam
al
on June 15, 2020 by guest. Protected by copyright.
http://gh.bmj.com
/B
MJ G
lob Health: first published as 10.1136/bm
jgh-2017-000543 on 6 March 2018. D
ownloaded from
Anstey Watkins J, et al. BMJ Glob Health 2018;3:e000543. doi:10.1136/bmjgh-2017-000543 9
BMJ Global Health
helped us to understand how and why each study’s inter-vention design and possible components of the interven-tion may have contributed to the study being effective or not. These theories were then used to help us search for higher-level frameworks (table 4) based on similar theo-ries to those described in table 3.
We found two frameworks to use in our analysis that brought together mechanisms and theories:
COM-B framework,40 designed for Behaviour change, describes the three domains of: Capability: physical skills, knowledge, behavioural regulation and mem-ory, attention and decision process; Opportunity: en-vironmental context, resources and social influences; Motivation: beliefs about consequences, optimism and beliefs about capabilities and reinforcement and emotion.
RFV framework,41 designed for the successful imple-mentation of telecommunications technology health interventions, describes the three domains of: Rela-tionships: with health workers and peer as a means of providing support for behavioural change, feedback and reinforcement; Fit: integration of mHealth into routine and its ease of use into the existing environ-ment and Visibility: to engage in information to medi-ate and motivate self-management tasks and enable enhanced awareness.
COM-B domains provide a framework for under-standing patient behaviours. RFV domains provide a framework to understand how mHealth interventions have been designed to be implemented. The domains of these frameworks include theory relevant to the mech-anisms of action of the interventions in our review. We used these theoretical frameworks to understand what mechanisms, in what context will result in a behaviour change.
The Realist and Meta-Narrative Evidence Syntheses (RAMESES I) reporting standards for realist reviews were followed.33
resultssearch resultsThe process of screening and paper selection resulted in 57 references (PRISMA diagram in figure 2). Of those, 29 papers were assessed for full text eligibility and a total of four studies were eligible for inclusion (table 2).
the studies and their designThe four papers included studies conducted in Paki-stan (Kamal),42 Honduras and Mexico (Piette),43 Kenya (Lester)44 and South Africa (Bobrow).45 The studies focused on specific diseases: stroke (Kamal), hyperten-sion (Bobrow and Piette) and HIV (Lester). Patients were all over 18 years of age. All four study designs were randomised controlled trials with control groups receiving usual standard care, all set in urban health-care settings. Two studies primarily measured clin-ical outcomes (Bobrow and Piette) and two measured
self-reported adherence (Lester and Kamal). The inter-ventions were designed to facilitate communication via mobile phones between the patient and health worker. All interventions involved medication adherence. Two studies had specific subgroup analyses; Piette’s subgroup was low-literate patients with information needs for how to manage high blood pressure; Bobrow’s subgroup was patients over 55 years of age who had good control of blood pressure at the baseline measure and patients on treatment for hypertension for less than 10 years.
The intervention design in three studies was text messages sent weekly to patients (Lester, Bobrow, Kamal). The other study’s intervention was weekly auto-mated phone calls (Piette) (table 2). In two studies, data submitted by patients were interpreted using an automated system (Kamal and Bobrow), whereas in the other two studies, there was engagement with a real health worker (Piette and Lester). In Bobrow’s study, there was an option for the patient to use their phones to initiate a free to user ‘Please Call Me’ request. A health worker did not actually phone the patient back. There was no verbal communication in this instance; instead, this request generated an automated series of responses from the text message delivery system. This design offered trial participants several options including cancelling or moving an appointment, and changing the timing and language of the incoming text messages, whereas in Bobrow’s and Kamal’s studies, the participants predefined when they wanted to receive the message to their phone. This gave them patient choice, as the other studies contacted the patient at a time decided on by the intervention team. In the case of Kamal’s study, when the intervention was described as automated, we had to assume that a health worker was reading the patient’s responses and taking action, especially when the patient had a problem. This was not made explicit in the paper and so it was unclear what the health worker’s role was.
Two studies (Lester and Bobrow), used standardised text messages whereas the other two (Kamal and Piette), used customised messages for each patient (table 2). In Kamal’s study, each patient received tailored text messages with information content about the patients’ individual prescriptions. Likewise, in Piette’s study, the automated calls could be tailored as patients were directly asked to provide blood pressure measurements by using the home monitoring kit, in addition to the patient’s mobile phone. In Lester’s study, they used very indirect ways of assisting patients to monitor as the content of the text messages were aimed at enquiring how they were feeling, without any reference to HIV, whereas in Bobrow’s trial, the information messages did include the words related to the patient’s chronic disease: ‘high blood pills’.
Intervention components and their intended purposeIntervention components were similar across the studies, but the combinations of varying components used were different. Intervention components included reminders, patient observation of health state, motivational
on June 15, 2020 by guest. Protected by copyright.
http://gh.bmj.com
/B
MJ G
lob Health: first published as 10.1136/bm
jgh-2017-000543 on 6 March 2018. D
ownloaded from
10 Anstey Watkins J, et al. BMJ Glob Health 2018;3:e000543. doi:10.1136/bmjgh-2017-000543
BMJ Global Health
Tab
le 4
M
echa
nism
s of
cha
nge
(RFV
and
CO
M-B
mec
hani
sms)
evi
den
t in
all
four
incl
uded
em
piri
cal s
tud
ies
on m
Hea
lth in
Ken
ya, S
outh
Afr
ica,
Hon
dur
as a
nd M
exic
o an
d P
akis
tan
RFV
and
CO
M-B
mec
hani
sms
by
stud
y au
tho
r
Inte
rven
tio
n co
mp
one
nts
used
in
eac
h st
udy
Rel
atio
nshi
ps
Fit
Vis
ibili
tyC
apab
ility
Op
po
rtun
ity
Mo
tiva
tio
n
Lest
er e
t al
44 (K
enya
)P
rovi
sion
of
sup
por
t co
mm
unic
atio
n,
Ob
serv
atio
n,
Rem
ind
ers
Wee
kly
com
mun
icat
ion
bet
wee
n p
atie
nt a
nd
nurs
e. Im
por
tanc
e of
p
erso
nal c
onta
ct fo
r m
otiv
atio
n.
Text
mes
sage
sup
por
t se
rvic
e an
d r
emin
der
m
essa
ge a
s ad
diti
on
to e
very
day
rou
tine
with
pat
ient
s w
ho w
ere
alre
ady
pho
ne u
sers
.
Rec
eivi
ng t
he r
emin
der
to
self-
rep
ort
adhe
renc
e to
H
IV t
reat
men
t in
crea
ses
the
awar
enes
s of
the
dis
ease
.
Not
ap
plic
able
If ill
itera
te, t
he H
IV p
atie
nt
mus
t b
e as
sist
ed b
y lit
erat
e p
artn
er (i
f will
ing
to d
iscl
ose
thei
r st
atus
/and
tha
t th
ey
are
on lo
ng t
erm
tre
atm
ent).
R
esp
ond
with
tha
t th
ey
are
doi
ng w
ell o
r ha
ve a
p
rob
lem
to
the
nurs
es.
Reg
ular
co
mm
unic
atio
n w
ith h
ealth
wor
ker
mot
ivat
es p
atie
nt t
o re
mem
ber
to
take
the
ir tr
eatm
ent
and
kee
p o
n to
p o
f the
ir d
isea
se.
Bob
row
et
al45
(Sou
th A
fric
a)R
emin
der
s,
Targ
eted
act
ions
Hea
lth w
orke
r re
latio
nshi
p w
ith
pat
ient
. Jus
t b
y re
ceiv
ing
the
mes
sage
giv
es
the
pat
ient
feel
ing
som
eone
car
es.
Text
mes
sage
as
add
ition
al r
emin
der
sy
stem
. Fun
ctio
n to
re
sche
dul
e ap
poi
ntm
ent
auto
mat
ical
ly v
ia fr
ee
retu
rn t
ext
mes
sage
.
Follo
w a
dhe
renc
e ad
vice
fr
om t
he t
ext
mes
sage
. Tex
t m
essa
ge in
crea
sed
dis
ease
aw
aren
ess,
pro
vid
ed t
ips
for
heal
th a
nd h
elp
ed t
o d
evel
op a
nd r
einf
orce
mor
e ro
bus
t re
min
der
sys
tem
s.
As
abov
e—b
e co
mfo
rtab
le w
ith
the
tech
nolo
gy t
o ac
cess
and
rea
d t
ext-
m
essa
ges.
Op
por
tuni
ty t
o m
ake
chan
ges
to c
linic
tim
es.
Mot
ivat
ion
to r
ead
th
e co
nten
t of
the
m
essa
ges
and
act
on
it: a
dhe
re.
Pie
tte
et a
l43 (H
ond
uras
and
M
exic
o)O
bse
rvat
ion,
R
emin
der
s,
Mot
ivat
ing
educ
atio
n/ad
vice
in
form
atio
n
Rel
atio
nshi
p
bet
wee
n p
atie
nt a
nd
the
kit
sent
to
the
heal
th w
orke
r.
Fits
into
the
ir ev
eryd
ay
lives
bec
ause
the
ki
t is
at
hom
e an
d is
p
urp
osel
y in
tend
ed fo
r th
is in
terv
entio
n.
It re
qui
res
logg
ing
blo
od
pre
ssur
e ev
ery
day
, so
it is
ver
y re
gula
r an
d t
his
inte
ract
ion
mak
es t
he
hyp
erte
nsio
n vi
sib
le in
the
p
erso
n’s
life.
The
per
son
need
s to
und
erst
and
ho
w t
o us
e th
e ne
w t
echn
olog
y in
th
eir
hom
e an
d t
he
confi
den
ce t
o us
e th
e ki
t an
d k
eep
usi
ng it
. Th
ey m
ust
buy
-in
to
why
it is
hel
pin
g th
em
to s
elf-
man
age.
Allo
ws
for
the
opp
ortu
nity
fo
r so
cial
sup
por
t an
d a
ch
oice
to
actu
ally
rec
eive
th
e su
pp
ort.
Mot
ivat
es p
erso
n to
re
duc
e sa
lt in
take
—
pow
erfu
l mes
sage
s re
late
d t
o b
ehav
iour
s ch
ange
.
Kam
al e
t al
42 (P
akis
tan)
Rem
ind
ers,
M
otiv
atin
g ed
ucat
ion/
advi
ce
info
rmat
ion,
Pra
ise
and
en
cour
agem
ent
Pat
ient
mus
t re
spon
d
to t
he h
ealth
wor
ker
to in
form
the
m
they
hav
e ta
ken
thei
r m
edic
atio
n.
Pat
ient
rec
eive
s tim
ely,
cus
tom
ised
m
essa
ges.
The
resp
onse
is y
es o
r no
, so
is q
uick
and
will
no
t ta
ke m
uch
of t
he
pat
ient
’s t
ime.
By
havi
ng t
o ta
ke t
he
med
icat
ion
and
the
n ac
tion
this
by
send
ing
a m
essa
ge,
it cr
eate
s d
isea
se v
isib
ility
. Th
e in
form
atio
n m
essa
ges
‘tak
e tw
o se
rvin
gs o
f fru
it’
enco
urag
e th
e p
atie
nt
to m
ake
heal
thy
choi
ces
sinc
e th
ey a
re m
anag
ing/
pre
vent
ing
futu
re s
trok
es.
The
pat
ient
mus
t ha
ve t
he c
apab
ility
to
resp
ond
to
the
cue
to
actio
n.
The
pat
ient
mus
t ha
ve t
he o
pp
ortu
nity
to
rep
ly in
clud
ing
the
com
pre
hens
ion
or s
ocia
l su
pp
ort.
The
pat
ient
mus
t ha
ve t
he m
otiv
atio
n to
res
pon
d t
o th
e he
alth
wor
ker
each
d
ay.
Sum
mar
y of
mec
hani
sms
that
fire
to
caus
e/en
able
ch
ange
and
pat
ient
’s
enga
gem
ent
in c
hron
ic
dis
ease
mon
itorin
g
Inte
rven
tion
mus
t al
low
for
a re
latio
nshi
p t
o fo
rm b
etw
een
the
pat
ient
and
the
he
alth
wor
ker
to
pro
duc
e m
otiv
atio
n.
Inte
rven
tion
mus
t fit
wel
l in
the
pat
ient
's li
fe a
nd
thei
r in
div
idua
l nee
ds.
Inte
rven
tion
mak
es t
he
dis
ease
vis
ible
.P
atie
nt n
eed
per
sona
l ca
pab
ility
to
resp
ond
to
the
mec
hani
sms
of
chan
ge.
Pat
ient
is o
ffere
d t
he
opp
ortu
nity
for
sup
por
t w
hich
may
mot
ivat
e th
em.
Pat
ient
nee
ds
mot
ivat
ion
to s
elf-
mon
itor.
on June 15, 2020 by guest. Protected by copyright.
http://gh.bmj.com
/B
MJ G
lob Health: first published as 10.1136/bm
jgh-2017-000543 on 6 March 2018. D
ownloaded from
Anstey Watkins J, et al. BMJ Glob Health 2018;3:e000543. doi:10.1136/bmjgh-2017-000543 11
BMJ Global Health
education/advice, provision of support communica-tion, targeted actions and praise and encouragement. Given the extracted results in table 2, in table 3, we then describe how each intervention’s components relate to theory. We found this midway process useful in helping to define the programme theory.
All except Bobrow’s study required a response from the patient. Instead, the two information groups received information-only messages and the second intervention group also had to the option to interact with the auto-mated message, if desired.
Programme theory: how the intervention was intended to workOnly Bobrow’s study provided a clear statement of the programme theory and how the authors intended their intervention to work (described in full, in another publi-cation46). It was supported by a process evaluation of the study in a paper by Leon.47 The other studies mentioned the theories on which they based their intervention design but did not explicitly explain why they felt their intervention worked. For Kamal and Lester’s studies, we used related published study protocols or supplementary
material to work out their programme theory. The programme theories are reported in table 2.
Programme theory summary from across the four included studies:
All four studies used weekly messages/calls. This was intended to make the patient feel cared for. These in-cluded non-health related information such as ‘Hap-py Birthday’ messages (Bobrow).
In three studies (Lester, Piette and Kamal), a non-re-sponse from the patient or the reporting of poor blood pressure readings triggered a health worker to follow-up the patient. This process of follow-up acted as incentive to the patient to always respond in the specified period (Piette).
In two studies (Bobrow and Kamal), patients were asked, when they wanted to receive the message at a time of day that was relevant to them. This prompted the patient to take action as they had chosen to re-ceive the communication at that specified time.
In Piette’s study, patients had to take their blood pres-sure measurements after a cue to action. The other three studies did not require the patient to take clin-ical measurements as home-monitoring kit was not
Figure 2 PRISMA.
on June 15, 2020 by guest. Protected by copyright.
http://gh.bmj.com
/B
MJ G
lob Health: first published as 10.1136/bm
jgh-2017-000543 on 6 March 2018. D
ownloaded from
12 Anstey Watkins J, et al. BMJ Glob Health 2018;3:e000543. doi:10.1136/bmjgh-2017-000543
BMJ Global Health
provided. Also, in Piette’s study, the patient’s report was also sent to a predefined treatment supporter. This was intended to make the patient feel they were accountable to somebody who cares for them. This was the only study to use an external person beyond the patient-health worker.
Two studies (Piette and Kamal) customised the mes-sages to ensure the patient felt more satisfied and to prompt them into action.
In two studies, the text messages required simple quick responses such as ‘Yes or No’ (Kamal) or ‘I am fine’ (Lester). These responses were appropriate as the patient did not have to spend much time replying.
Mechanisms of actionAll four studies used the domains of RFV and COM-B as mechanisms of change to some degree (table 4).
rFVRelationshipThere needs to be personal contact within the relation-ship between the patient and health worker to motivate the patient or the text messages need to be tailored to make the patient feel like someone cares. Fit: The level of how well the mHealth intervention fits into the patient’s daily life is crucial to how they will respond. Visibility: It appears that visibility acts as a mechanism in the studies because the mHealth component’s purpose is to improve monitoring and to therefore make the disease more visible in the patients’ lives, at least once a week.
CoM-bCapability, Opportunity, Motivation=BehaviourThe interpersonal relationship between the patient and health worker created when the mHealth interven-tion is not automated improves the patient’s motiva-tion to feel capable of responding or adhering to their treatment regimen. When the patient feels cared for by the real person who is involved, this leads to change in the patients’ behaviour. When the patient is offered the opportunity to be given information that provides education and advice, this may allow them to engage with the information. This in turn may improve the way they manage their disease, as they feel more capable of doing so. Even automated communication can motivate patients to engage.
In all four studies, the patients engage with the inter-vention components (mHealth resources) by using their own reasoning. Their reasoning is made up of their own motivation and empowerment to respond to the resources. It is this engagement that leads to change and thus improvements in the study’s primary and/or secondary outcomes (table 2, final column). Table 4 summarises the combinations of mechanisms for each intervention.
In Lester’s study, the mHealth components of provi-sion of supportive communication, self-observation and reminders prompted engagement with a health worker
(relationship). This personal contact with the nurse resulted in the patient being motivated to collect their medication from the clinic; otherwise a nurse will check-up on them. This personal contact with the nurse and feeling more supported resulted in the patient being motivated to take their treatment (table 4) and keep on top of their disease management. However, this did not lead to improved adherence in treatment.
In Bobrow’s study, the mHealth components were reminders, opportunity to make changes to appointment times so the patient was able to fit the intervention into their daily life. The text messages increase disease aware-ness and thus its visibility. The mHealth intervention allowed the patient to feel cared for by the nurse and to take more control of their management by having the option to change appointments from their phones. This did not lead to a decrease in blood pressure; however, adherence to medication improved.
In Piette’s study, reminders and motivating educa-tion/advice information prompted engagement with a health worker and treatment supporter (relationship). The patient felt they had empathetic support, especially when the patient had problems. Disease management was made more visible as the patient had to log onto the home-monitoring kit every day. Information messages prompted change in lifestyle. This regular interaction resulted in the opportunity to engage with social support when they needed it. By reading powerful practical messages, this motivated them to reduce salt intake for example. This led to a decrease in blood pressure, partic-ularly in the subgroup.
In Kamal’s study the reminders and motivating educa-tion/advice information were tailored to the individual patients making the patient feel they had a relationship with their health worker, even though the intervention did not have any direct patient–health worker commu-nication. By having to respond to the health worker, this resulted in the quick ‘Yes’ or ‘No’ response message becoming routinised and did not take much of the patient’s time. This led to improved adherence to medi-cation and a slight reduction in diastolic blood pressure. Table 4 summarises which combinations of mechanisms help to determine what is the outcome of the mHealth intervention.
study outcomesAll studies, except Bobrow’s study, achieved their primary outcome. Two studies (Piette and Kamal) had short follow-up periods of 6–8 weeks, respectively. Piette’s inter-vention was effective in decreasing blood pressure and Kamal’s intervention was effective in increasing adher-ence to personalised stroke medication. Two studies (Lester and Bobrow) collected follow-up data at 6 months and 12 months. Lester’s study showed effectiveness of self-reported HIV adherence at both 6 and 12 months and improved viral load suppression at 12 months. Bobrow’s study found a non-significant reduction in systolic blood pressure control compared with usual care at both 6 and
on June 15, 2020 by guest. Protected by copyright.
http://gh.bmj.com
/B
MJ G
lob Health: first published as 10.1136/bm
jgh-2017-000543 on 6 March 2018. D
ownloaded from
Anstey Watkins J, et al. BMJ Glob Health 2018;3:e000543. doi:10.1136/bmjgh-2017-000543 13
BMJ Global Health
12 months. There was significant improvement in their secondary outcome of adherence to collecting medica-tion from the clinic.
Variation in outcomes by context-related patient factorsIn Bobrow’s and Lester’s studies, the interventions were more effective for patients early in their disease trajectory. The interventions are likely to have given the patients opportunity to establish disease management routines. In Piette’s study, the intervention was designed to meet the needs of patients with low literacy and the intervention did have greater effect on this subgroup of patients. The intervention is likely to have increased patient knowledge (from a low level) and so increased opportunity for the patient to engage with managing their disease. The inter-vention is likely to have challenged beliefs about disease consequences and about their capability to manage their disease leading to increased motivation. Bobrow’s study found that patients who benefited the most from the mHealth intervention were those with high personal stress caused by multiple psychosocial stressors. The intervention is likely to have provided a structure to their disease management that was lacking in other aspects of their life, increasing their motivation to engage with disease management yet without intruding into other aspects of life. We acknowledge there are certain subgroups that respond particularly well to mHealth interventions (and were under subanalyses within the studies) and where their monitoring improves and why this may be so. Patients with low literacy/high informa-tion needs who require knowledge and support (any information is better than nothing) and older patients who have good blood pressure control and/or have been diagnosed within the past 10 years.
dIsCussIonAll the interventions were effective in terms of improving adherence to monitoring of the long-term condition for which they were designed (HIV, hypertension and stroke management). The interventions in the four studies included different combinations of intervention compo-nents but each one included a reminder. We were able to map all intervention components to RFV and/or COM-B mechanisms. Some interventions focused more on the relationship aspect of care provision and others on moti-vating the patient. Subgroup analysis within three studies suggests that contextual factors moderate the impact of the interventions. In the studies that we reviewed, these factors were patient related: low literacy, high personal stress and time since diagnosis. We found no evidence of moderation by non-patient contextual factors.
Our findings are consistent with those of other inter-national reviews (some within high-income settings). Three systematic reviews found reminders for disease monitoring improve healthcare processes but with education through voice call and text messages, both health outcomes and care processes improve.18 22 23 A
realist review of mobile phone-based health interven-tions for non-communicable disease management in sub-Saharan Africa48 found that patient related contex-tual factors influenced the impact of interventions. Similarly, text messaging used to improve adherence to antiretroviral therapy had more effect for those with at least primary education.49 Taking account of sociocul-tural factors has been shown to influence scale-up and sustainability of mHealth interventions.50 Reviews have found variation in effect depending on how mechanisms were operationalised. For example, weekly reminders worked better than daily reminders, and messages had more impact when they were linked with patient–health worker interaction.17 27 49 51–53 In our review, we were unable to identify variation in impact related to how the mechanisms were operationalised. Our review found no studies where patients received test results or instructions to adjust medication. In high-income countries, interventions have included these aspects of care.18 22 23 Most reviews of use of digital communi-cation with patients in sub-Saharan Africa, comment on the problems of implementation of the communi-cation system, particularly technical and maintenance issues.26–28 54–58 In our included studies, implementation issues had been resolved.
strengths and limitationsA realist approach was methodologically appropriate for our research question. However, we were limited by the design and reporting of the included studies. We do not know whether all of the study authors were aware if their intervention designs included domains of RFV and COM-B as mechanisms of change. Among our included studies, there were no interventions of sufficient simi-larity to allow comparison of outcomes when used in different contexts. Within each study, the authors did not describe variation in the context in which the inter-vention was delivered. This limits our ability to critically examine the programme theory and could weaken the explanatory power of our conclusions.
The lack of detail in the included study interventions may mean we incorrectly interpreted how and why the intervention worked. Only one of the three studies included a process evaluation. Interviews with the study authors would have clarified their programme theory for their intervention. This would possibly deepen our anal-ysis by accounting for the socioecological dimensions of behaviour rather than being individually focused. The studies had limited follow-up over time, so sustainability of intervention use and sustainability of effect is unclear. The lack of evidence on sustainability is recognised as a serious problem in mHealth pilots in low-resource settings.59 The only contextual factors we identified were those measured as patient characteristics and used in subgroup analyses. Also, we may have missed papers for inclusion if we had included non-English papers and extended our search to nursing databases.
on June 15, 2020 by guest. Protected by copyright.
http://gh.bmj.com
/B
MJ G
lob Health: first published as 10.1136/bm
jgh-2017-000543 on 6 March 2018. D
ownloaded from
14 Anstey Watkins J, et al. BMJ Glob Health 2018;3:e000543. doi:10.1136/bmjgh-2017-000543
BMJ Global Health
Policy and practice implications and future researchOur review suggests there is potential for improving the monitoring of chronic disease in LMICs using two-way digital communication, although close connection with local care provision is important. Behavioural theory can guide the design of mHealth interventions aimed at changing health behaviours.60 Wong et al33 states that realist reviews reveal ‘what policy makers or practitioners might put in place to change the context or provide resources in such a way as to most likely trigger the right mechanism(s) to produce the desired outcome’. It is important to target interventions where impact is most likely, for example, when a patient has more recently been diagnosed. Other contexts which information based mHealth may contribute are when patients have low literacy and low knowledge about their disease. Patients who are under stressful life situations such as poverty, family bereavement or managing multiple long-term conditions may also be amenable and respond well to mHealth support.
Further research is needed on the tailoring of mHealth interventions to the needs and sociocultural context of the patient. For example, the *STAR2D study61 underway has found that tailoring dietary advice to patient context is appropriate.46
ConClusIonAlthough chronic disease management is a major burden on healthcare providers in low-resource contexts, there is very little evidence for the use of mHealth for improving chronic disease monitoring in these settings and what exists is mainly on medication adherence. The interven-tions that have been evaluated mapped to established behaviour theory though this was not always explicit in their description of their mHealth design. There was little evidence of how context moderated the effect of the intervention, except for contextual factors reflected in patient characteristics such as low literacy, under stress or being early in the disease trajectory.
Acknowledgements We thank Samantha Johnson, Medical Academic Librarian, University of Warwick, UK for support with the review’s search strategy; Professor Geoff Wong, University of Oxford, UK for his realist knowledge; Professor Donald Cole, University of Toronto, Canada for his advice early in the review and the journal reviewers for their comments.
Contributors JAW led the review, the analysis and drafted the paper with support throughout from FG, who also helped define the scope of the review. CH assisted with methodological expertise at the start of the review. JG and FXG-O read various iterations during the review process, giving guidance and advice. KD quality checked the extracted data. All authors provided feedback on the drafts and read and approved the final manuscript.
Funding This review contributed to a PhD thesis called ‘The Vutivi study’ (http:// webcat. warwick. ac. uk/ record= b3045268~ S1), fully funded by the UK’s Economic and Social Research Council (ESRC ES/J500203/1). GE Healthcare Ltd. contributed funding towards research fieldwork costs for the Vutivi Study and played no role in shaping the research for the PhD.
Competing interests None declared.
Patient consent Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
data sharing statement All data will be made available on request.
open Access This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http:// creativecommons. org/ licenses/ by/ 4. 0/
© Article author(s) (or their employer(s) unless otherwise stated in the text of the article) 2018. All rights reserved. No commercial use is permitted unless otherwise expressly granted.
RefeRences 1. Ad-G A, Agyemang C. Chronic Non-communicable Diseases in Low
and Middle-income Countries: CABI, 2015. 2. Beran D, Chappuis F, Cattacin S, et al. The need to focus on primary
health care for chronic diseases. Lancet Diabetes Endocrinol 2016;4:731–2.
3. United Nations. Sustainable Development Goals: 17 Goals to Transform our World United Nations 2016. http ://w ww.u n.org/ sustainabledev elo pmen t/ health/ (accessed 10 Jan 2017).
4. Larsen PD, Lubkin IM. Chronic illness: Impact and intervention: Jones & Bartlett Learning, 2009.
5. Schulman-Green D, Jaser S, Martin F, et al. Processes of self-management in chronic illness. J Nurs Scholarsh 2012;44:136–44.
6. Shippee ND, Shah ND, May CR, et al. Cumulative complexity: a functional, patient-centered model of patient complexity can improve research and practice. J Clin Epidemiol 2012;65:1041–51.
7. Kadirvelu A, Sadasivan S, Ng SH, Sh N. Social support in type II diabetes care: a case of too little, too late. Diabetes Metab Syndr Obes 2012;5:407.
8. Glasziou P, Irwig L, Mant D. Monitoring in chronic disease: a rational approach. BMJ 2005;330:644–8.
9. Hamine S, Gerth-Guyette E, Faulx D, et al. Impact of mHealth chronic disease management on treatment adherence and patient outcomes: a systematic review. J Med Internet Res 2015;17:e52.
10. Gaziano TA, Pagidipati N. Scaling up chronic disease prevention interventions in lower- and middle-income countries. Annu Rev Public Health 2013;34:317–35.
11. Mahomed OH, Asmall S. Development and implementation of an integrated chronic disease model in South Africa: lessons in the management of change through improving the quality of clinical practice. Int J Integr Care 2015;15.
12. Stephani V, Opoku D, Quentin W. A systematic review of randomized controlled trials of mHealth interventions against non-communicable diseases in developing countries. BMC Public Health 2016;16:572.
13. Beratarrechea A, Moyano D, Irazola V, et al. mHealth interventions to counter noncommunicable diseases in developing countries: still an uncertain promise. Cardiol Clin 2017;35:13–30.
14. Anstey Watkins JOT, Goudge J, Gómez-Olivé FX, et al. Mobile phone use among patients and health workers to enhance primary healthcare: A qualitative study in rural South Africa. Soc Sci Med 2018;198:139–47.
15. Odendaal WA, Goudge J, Griffiths F, et al. Healthcare workers’ perceptions and experience on using mHealth technologies to deliver primary healthcare services: qualitative evidence synthesis. Cochrane Database Syst Rev 2015;2010:1–14.
16. Vasudevan L, Zeller K, Labrique A, et al. Mobile health. Digital health: Springer, 2018:15–25.
17. Bärnighausen T, Chaiyachati K, Chimbindi N, et al. Interventions to increase antiretroviral adherence in sub-Saharan Africa: a systematic review of evaluation studies. Lancet Infect Dis 2011;11:942–51.
18. Cole-Lewis H, Kershaw T. Text messaging as a tool for behavior change in disease prevention and management. Epidemiol Rev 2010;32:56–69.
19. Gurol-Urganci I, de Jongh T, Vodopivec-Jamsek V, et al. Mobile phone messaging reminders for attendance at healthcare appointments. Cochrane Database Syst Rev 2013:CD007458.
20. Yasmin F, Banu B, Zakir SM, et al. Positive influence of short message service and voice call interventions on adherence and health outcomes in case of chronic disease care: a systematic review. BMC Med Inform Decis Mak 2016;16:46.
21. Mbuagbaw L, Mursleen S, Lytvyn L, et al. Mobile phone text messaging interventions for HIV and other chronic diseases: an overview of systematic reviews and framework for evidence transfer. BMC Health Serv Res 2015;15:33.
22. Krishna S, Boren SA, Balas EA. Healthcare via cell phones: a systematic review. Telemed J E Health 2009;15:231–40.
on June 15, 2020 by guest. Protected by copyright.
http://gh.bmj.com
/B
MJ G
lob Health: first published as 10.1136/bm
jgh-2017-000543 on 6 March 2018. D
ownloaded from
Anstey Watkins J, et al. BMJ Glob Health 2018;3:e000543. doi:10.1136/bmjgh-2017-000543 15
BMJ Global Health
23. Free C, Phillips G, Galli L, et al. The effectiveness of mobile-health technology-based health behaviour change or disease management interventions for health care consumers: a systematic review. PLoS Med 2013;10:e1001362.
24. van Velthoven MH, Brusamento S, Majeed A, et al. Scope and effectiveness of mobile phone messaging for HIV/AIDS care: a systematic review. Psychol Health Med 2013;18:182–202.
25. Fjeldsoe BS, Marshall AL, Miller YD. Behavior change interventions delivered by mobile telephone short-message service. Am J Prev Med 2009;36:165–73.
26. Bloomfield GS, Vedanthan R, Vasudevan L, et al. Mobile health for non-communicable diseases in Sub-Saharan Africa: a systematic review of the literature and strategic framework for research. Global Health 2014;10:49.
27. Catalani C, Philbrick W, Fraser H, et al. mHealth for HIV treatment & prevention: a systematic review of the literature. Open AIDS J 2013;7:17–41.
28. Chib A, van Velthoven MH, Car J. mHealth adoption in low-resource environments: a review of the use of mobile healthcare in developing countries. J Health Commun 2015;20:4–34.
29. Greenhalgh T, Wong G, Westhorp G, et al. Protocol--realist and meta-narrative evidence synthesis: evolving standards (RAMESES). BMC Med Res Methodol 2011;11:115.
30. Otte-Trojel T, Wong G. Going beyond systematic reviews: realist and meta-narrative reviews. Stud Health Technol Inform 2016;222:275–87.
31. Wong G, Pawson R, Owen L. Policy guidance on threats to legislative interventions in public health: a realist synthesis. BMC Public Health 2011;11:222.
32. Pawson R, Greenhalgh T, Harvey G, et al. Realist review--a new method of systematic review designed for complex policy interventions. J Health Serv Res Policy 2005;10 Suppl 1:21–34.
33. Wong G, Greenhalgh T, Westhorp G, et al. RAMESES publication standards: realist syntheses. J Adv Nurs 2013;69:1005–22.
34. Rycroft-Malone J, McCormack B, Hutchinson AM, et al. Realist synthesis: illustrating the method for implementation research. Implement Sci 2012;7:33.
35. Michie S, Thomas J, Johnston M, et al. The human behaviour-change project: harnessing the power of artificial intelligence and machine learning for evidence synthesis and interpretation. Implement Sci 2017;12:121.
36. Anstey-Watkins. Understanding the potential role for appropriate digital technological solutions in the innovation of health system design, implementation and normalisation in rural South Africa for both patients and health-workers: A critical exploratory analysis: University of Warwick, 2016. PhD Thesis.
37. Lacouture A, Breton E, Guichard A, et al. The concept of mechanism from a realist approach: a scoping review to facilitate its operationalization in public health program evaluation. Implement Sci 2015;10:1:1.
38. Pawson R, Tilley N. Realistic evaluation: Sage, 1997. 39. Dalkin SM, Greenhalgh J, Jones D, et al. What's in a mechanism?
Development of a key concept in realist evaluation. Implement Sci 2015;10:49.
40. Michie S, Richardson M, Johnston M, et al. The behavior change technique taxonomy (v1) of 93 hierarchically clustered techniques: building an international consensus for the reporting of behavior change interventions. Ann Behav Med 2013;46:81–95.
41. Vassilev I, Rowsell A, Pope C, et al. Assessing the implementability of telehealth interventions for self-management support: a realist review. Implement Sci 2015;10:59.
42. Kamal AK, Shaikh Q, Pasha O, et al. A randomized controlled behavioral intervention trial to improve medication adherence in adult stroke patients with prescription tailored Short Messaging Service (SMS)-SMS4Stroke study. BMC Neurol 2015;15:212.
43. Piette JD, Datwani H, Gaudioso S, et al. Hypertension management using mobile technology and home blood pressure monitoring: results of a randomized trial in two low/middle-income countries. Telemed J E Health 2012;18:613–20.
44. Lester RT, Ritvo P, Mills EJ, et al. Effects of a mobile phone short message service on antiretroviral treatment adherence in Kenya (WelTel Kenya1): a randomised trial. Lancet 2010;376:1838–45.
45. Bobrow K, Farmer AJ, Springer D, et al. Mobile phone text messages to support treatment adherence in adults with high blood pressure (SMS-Text Adherence Support [StAR]): a single-blind, randomized trial. Circulation 2016;133:CIRCULATIONAHA.115.017530.
46. Bobrow K, Farmer A, Cishe N, et al. Using the medical research council framework for development and evaluation of complex
interventions in a low resource setting to develop a theory-based treatment support intervention delivered via SMS text message to improve blood pressure control. BMC Health Serv Res 2018;18:33.
47. Leon N, Surender R, Bobrow K, et al. Improving treatment adherence for blood pressure lowering via mobile phone SMS-messages in South Africa: a qualitative evaluation of the SMS-text Adherence SuppoRt (StAR) trial. BMC Fam Pract 2015;16:80.
48. Opoku D, Stephani V, Quentin W. A realist review of mobile phone-based health interventions for non-communicable disease management in sub-Saharan Africa. BMC Med 2017;15:24.
49. Mbuagbaw L, van der Kop ML, Lester RT, et al. Mobile phone text messages for improving adherence to antiretroviral therapy (ART): an individual patient data meta-analysis of randomised trials. BMJ Open 2013;3:003950.
50. Greenhalgh T, Wherton J, Papoutsi C, et al. Beyond adoption: a new framework for theorizing and evaluating nonadoption, abandonment, and challenges to the scale-up, spread, and sustainability of health and care technologies. J Med Internet Res 2017;19:e367.
51. Horvath T, Azman H, Kennedy GE, et al. Mobile phone text messaging for promoting adherence to antiretroviral therapy in patients with HIV infection (Review). Cochrane Database of Systematic Reviews 2012;2012:009756.
52. Pop-Eleches C, Thirumurthy H, Habyarimana JP, et al. Mobile phone technologies improve adherence to antiretroviral treatment in a resource-limited setting: a randomized controlled trial of text message reminders. AIDS 2011;25:825–34.
53. Lall D, Prabhakaran D. Organization of primary health care for diabetes and hypertension in high, low and middle income countries. Expert Rev Cardiovasc Ther 2014;12:987–95.
54. Aranda-Jan CB, Mohutsiwa-Dibe N, Loukanova S. Systematic review on what works, what does not work and why of implementation of mobile health (mHealth) projects in Africa. BMC Public Health 2014;14:188.
55. Leon N, Schneider H. MHealth4CBS in South Africa: a review of the role of mobile phone technology for monitoring and evaluation of community based health services: Medical Research Council of South Africa (MRC): Medical Research Council of South Africa (MRC). Health Systems Research Unit, 2012:1–37.
56. Källander K, Tibenderana JK, Akpogheneta OJ, et al. Mobile health (mHealth) approaches and lessons for increased performance and retention of community health workers in low- and middle-income countries: a review. J Med Internet Res 2013;15:e17.
57. Betjeman TJ, Soghoian SE, Foran MP. mHealth in Sub-Saharan Africa. Int J Telemed Appl 2013;6.
58. Gurman TA, Rubin SE, Roess AA. Effectiveness of mHealth behavior change communication interventions in developing countries: a systematic review of the literature. J Health Commun 2012;17 Suppl 1:82–104.
59. Sanner TA, Roland LK, Braa K. From pilot to scale: towards an mHealth typology for low-resource contexts. Health Policy Technol 2012;1:155–64.
60. Michie S, Yardley L, West R, et al. Developing and evaluating digital interventions to promote behavior change in health and health care: recommendations resulting from an international workshop. J Med Internet Res 2017;19:e232.
61. Farmer A. *STAR2D - Text-messaging for type 2 diabetes adherence support in SAA. 2017 https://www. phc. ox. ac. uk/ phctrials/ trial- portfolio/ star-d (accessed on 11 Dec 2017).
62. Northouse PG, Northouse LL. Health communication: strategies for health professionals: appleton and Lange, 1992.
63. Martin LR, Haskard-Zolnierek KB, DiMatteo MR. Book and media reviews. JAMA 2011;305:1.
64. Weinstein N. Human motivation and interpersonal relationships: Springer, 2014.
65. Petty R, Cacioppo JT. Communication and persuasion: central and peripheral routes to attitude change: Springer Science & Business Media, 2012.
66. Rosenstock IM. The health belief model and preventive health behavior. Health Educ Monogr 1974;2:354–86.
67. Ajzen I. The theory of planned behavior. Organ Behav Hum Decis Process 1991;50:179–211.
68. Plass JL, Moreno R, Brünken R. Cognitive load theory. Cambridge: Cambridge University Press, 2010.
69. Locke E, Latham G. Goal-setting theory. Buffalo: University of New York, 1994:159.
70. Bem DJ. Self-perception theory. New York: Academic Press Inc, 1973.
on June 15, 2020 by guest. Protected by copyright.
http://gh.bmj.com
/B
MJ G
lob Health: first published as 10.1136/bm
jgh-2017-000543 on 6 March 2018. D
ownloaded from