doi:10.1136/jech.2008.078725 Apr 2009;
2009;63;610-622; originally published online 21J Epidemiol Community Health S Michie, K Jochelson, W A Markham and C Bridle
effectiveness and theoretical frameworksinterventions: a review of intervention content, Low-income groups and behaviour change
http://jech.bmj.com/cgi/content/full/63/8/610Updated information and services can be found at:
These include:
References
http://jech.bmj.com/cgi/content/full/63/8/610#BIBL
This article cites 36 articles, 10 of which can be accessed free at:
Rapid responses http://jech.bmj.com/cgi/eletter-submit/63/8/610
You can respond to this article at:
serviceEmail alerting
the top right corner of the article Receive free email alerts when new articles cite this article - sign up in the box at
Topic collections
(3570 articles) Smoking � (6585 articles) Health promotion � (6734 articles) Health education �
(4255 articles) Health service research � (2995 articles) Smoking and tobacco �
Articles on similar topics can be found in the following collections
Notes
http://journals.bmj.com/cgi/reprintformTo order reprints of this article go to:
http://journals.bmj.com/subscriptions/ go to: Journal of Epidemiology and Community HealthTo subscribe to
on 16 July 2009 jech.bmj.comDownloaded from
Low-income groups and behaviour changeinterventions: a review of intervention content,effectiveness and theoretical frameworks
S Michie,1 K Jochelson,2 W A Markham,3 C Bridle4
1 Research Department ofClinical, Education and HealthPsychology, University CollegeLondon, London, UK; 2 Equalityand Human Rights Commission,London, UK; 3 School of Healthand Social Studies, University ofWarwick, Coventry, UK;4 Clinical Trials Unit, WarwickMedical School, University ofWarwick, Coventry, UK
Correspondence to:Professor S Michie, ResearchDepartment of Clinical,Education and HealthPsychology, University CollegeLondon, 1–19 Torrington Place,London WC1E 7HB, UK;[email protected]
Accepted 2 March 2009
ABSTRACTBackground: Interventions to change health-relatedbehaviours have potential to increase health inequalities.Methods: This review investigated the effectiveness ofinterventions targeting low-income groups to reducesmoking or increase physical activity and/or healthy eating.Of 9766 papers identified by the search strategy, 13 met theinclusion criteria. Intervention content was coded intocomponent technique and theoretical basis, and examinedas a potential source of effect heterogeneity.Results: Interventions were heterogeneous, comprising4–19 techniques. Nine interventions had positive effects,seven resulted in no change and one had an adverseeffect. Effective interventions had a tendency to havefewer techniques than ineffective interventions, with noevidence for any technique being generally effective orineffective. Only six studies cited theory relative tointervention development, with little information abouthow theory was used and no obvious association withintervention content or effect.Conclusion: This review shows that behaviour changeinterventions, particularly those with fewer techniques,can be effective in low-income groups, but highlights thelack of evidence to draw on in informing the design ofinterventions for disadvantaged groups.
Chronic diseases, such as cardiovascular disease,diabetes, cancer and respiratory disease, are a majorcause of death and disability worldwide. There isconsiderable evidence showing that quitting smok-ing, eating a healthier diet, not consuming exces-sive amounts of alcohol and exercising regularlycan have a major impact on reducing rates ofchronic illness1 2 Lower socioeconomic status (SES)and lower social status is associated with poorerhealth outcomes and less healthy behaviours.3–6 Forexample, in Great Britain, although smokingprevalence has declined across all socioeconomicgroups, 15% of managerial and professional groupssmoked compared with 29% of manual occupa-tional groups in 2006.7 While 30% of adults inmanagerial and professional groups eat the recom-mended five portions of fruit and vegetable a day,just 18% of adults in the routine and manualgroups do so.8 Only 25% of people in lowersocioeconomic groups participate in sports andexercise compared with about 50% of highersocioeconomic groups (although when occupa-tional activity is controlled for, activity levels aresimilar).8 9 The adult routine and manual group isestimated at about 15 million people; about 4.3million smoke, 12.3 million eat less than five fruitand vegetable portions a day, and 7.5 million are
not physically active.10 Even a small percentagechange in behaviour in lower socioeconomic groupscould have a large impact on the health profile ofthe general population and on health costs.
The health promotion literature offers manytheories and techniques on behaviour change, butthus far there has been little research analysing theeffectiveness of particular component techniques,or of the effectiveness of techniques acrossdifferent groups. Literature reviews of particularhealth behaviours point to the paucity of data onthe impact of health promotion programmes onbehaviour change in poor and socially excludedgroups.11–13 Recent reviews also noted the lack ofinformation at review or meta-review level on thevariable effects of interventions on different socio-economic groups and on the impact of interven-tions on reducing health inequalities14–16 Albarracinet al show that the impact of interventions iscontingent on gender, age, ethnicity and otherpopulation-specific factors, suggesting that genericinterventions cannot be applied across populationswith confidence that they will be effective.16
There is clear evidence that people from dis-advantaged backgrounds are less successful inachieving behaviour change following participationin formal programmes such as smokers’ clinics.17 18
However, this does not necessarily mean thatthose programmes were less effective; it may bethat those from disadvantaged backgrounds beganwith a lower chance of success because of theirstarting levels of behaviour, and their physical and/or social environments undermine attempts atchange. There is consistent evidence that smokersin low socioeconomic groups are significantly morelikely to fail to quit smoking compared withsmokers in higher socioeconomic groups.19
Community-based programmes promoting healthyeating and physical activity have more difficultyrecruiting participants from low socioeconomicgroups20 and find higher attrition rates amonglow-income participants.21
Reducing health inequalities depends on devel-oping interventions to increase healthy behavioursthat are differentially effective in favour of thosefrom disadvantaged backgrounds or that targetsocially disadvantaged groups. A recent review onbehaviour change drew attention to the lack ofreviews and primary studies investigating differ-ential effectiveness among social groups and thelack of research on the cost-effectiveness ofbehaviour change interventions.15 The few studiesthat have investigated the effectiveness of inter-ventions across socioeconomic groups have tended
Essay
610 J Epidemiol Community Health 2009;63:610–622. doi:10.1136/jech.2008.078725
on 16 July 2009 jech.bmj.comDownloaded from
to do this as post hoc comparisons and are underpowered to dothis, leading to equivocal results.
This review focuses on interventions specifically targeted atlow-income populations, one index of disadvantage. It includedthree targeted behaviours related to health: smoking, healthyeating and physical activity. These were selected as they arehighly associated with illness and death, eg, they constituted70% of the modifiable behaviours found to be associated withdeath in the United States in 2000.22 This review analysesintervention content into its component techniques; such adetailed description is necessary for evaluating effectiveness andfor understanding mechanisms of change.23 24 Interventions aredescribed using a reliable taxonomy of behaviour changetechniques.25
The increasing recognition that interventions to changebehaviour should draw on theories of behaviour and behaviourchange in their development26 is for three main reasons.24 First,interventions are likely to be more effective if they target causaldeterminants of behaviour and behaviour change. Second,theory-based interventions facilitate an understanding of whyparticular interventions work and thus provide a basis fordeveloping better interventions across different contexts,populations and behaviours. Third, theory can be advancedonly if interventions and evaluations are theoreticallyinformed.24 25 However, many studies do not make the linkbetween theoretical models, expected outcomes and the processof change explicit.
The aim of this review is to identify evidence for theeffectiveness of health behaviour interventions that target low-income groups, with the aim of reducing smoking andunhealthy eating or increasing physical activity. It focuses onthe component techniques of the interventions, the theoriesused to develop the interventions, and considers associationsbetween theory and intervention content, and between inter-vention content and effect. It should be stressed that thisapproach is unable to comment on differential effectivenessacross social groups, only on whether there was any evidence ofeffect in interventions targeting low-income groups.
METHOD
Search strategyWe searched 21 electronic databases (January 1995 to September2006) using search terms related to a low-income population(eg, socioeconomic status, deprivation, disadvantaged, income)and three behaviours related to health: smoking cessation,healthy eating and physical activity (see Appendix A fordatabases and Appendix B for example search strategies). Inaddition, we approached 24 experts in the health inequalitiesfield enquiring about potential studies missed by our electronicsearch strategy, and we checked the bibliographies of allincluded studies.
ScreeningTitles and abstracts were screened against the inclusion criteriaby one reviewer, and a second reviewer independently checked arandom 10% of the search results. Full-text papers of referencesthat could not be excluded were ordered, and divided equallyamong two reviewers for assessment against the inclusioncriteria. For each reviewer, all papers initially selected forinclusion, and a random 10% of papers initially excluded, wereindependently assessed by the other reviewer. Inter-rateragreement was very good, kappa = 0.81.
Eligibility criteriaThe eligibility criteria for inclusion in this study were:1. Population: non-clinical, general population adults
(18+ years) from a low-income group.
2. Interventions: any interventions promoting smokingcessation, healthy eating and/or physical activity targetedat low-income groups.
3. Outcomes: behavioural outcomes relevant to the interven-tion target, ie, smoking cessation and increased healthyeating and physical activity.
4. Date: published after January 1995.
5. Language: published in the English language.
6. Methodological criteria: concurrent control, with or with-out random allocation. This therefore excludes reviews.
Data extractionA reviewer extracted data from the primary studies and asecond reviewer checked all papers for accuracy; discrepancieswere resolved through discussion. The interventions were codedby study design, country, target behaviour, type of participant,type of theory cited by the authors as informing theintervention, type of intervention and intervention effect.Intervention content was analysed into component techniques,using a reliable published taxonomy of 26 techniques,35 but twoadditional techniques were also identified. When interventionstargeted more than one behaviour, the techniques and resultswere recorded for each behaviour.
Data analysisOwing to the heterogeneity in intervention content and design,and the small sample, meta-analysis was not appropriate. Thenumber of techniques in effective and ineffective interventionswere tested statistically using independent t-tests.
RESULTS
Available evidenceAs figure 1 illustrates, the search strategy identified 9766references to potentially relevant studies, of which 1468 wereduplicates. We screened 8298 distinct references, excluding 8025primarily because the study populations were not of lowincome. We ordered full-text copies of the remaining 273references, and assessed 264 (nine could not be obtained, seeAppendix C), of which 238 were excluded primarily for failingto satisfy the population criterion. A total of 13 studies,
Figure 1 Flow of study papers through the review.
Essay
J Epidemiol Community Health 2009;63:610–622. doi:10.1136/jech.2008.078725 611
on 16 July 2009 jech.bmj.comDownloaded from
reporting 17 comparisons of interest, were included in thereview (see table 1).
Of the 13 studies included in the review, three wereconducted in the United Kingdom,31 38 39 one in Canada,34 eightin the USA,27–30 32 33 40–43 and one in the Netherlands.44 In total,seven, six and four studies developed interventions to promotesmoking cessation, healthy eating and physical activity,respectively, with three studies targeting more than onebehaviour.32–34
Eight studies were randomised controlled trials(RCTs),28 30–32 38 39 42 43 and five studies evaluated interventionsamong a non-randomised cohort with concurrent con-trol.33 34 40 41 44 Sample sizes ranged from 15 to more than 2000,although most were between 200 and 1000, with an averageapproaching 500. Effect data were extracted for the final follow-up, which ranged from less than 6 months28 to between 6 and12 months,30 32 33 39 42 43 to 12 months or more.31 34 38 40 41 44 Noneof the studies investigated cost-effectiveness.
Intervention techniquesInterventions were very heterogeneous, incorporating anythingfrom 4 to 19 techniques. Figure 2 shows the number ofinterventions incorporating each technique (multiple beha-viours targeted by one study intervention are counted asseparate interventions). Those used most frequently (in at least9 of the 17 interventions) were: providing general information;providing information about the consequences of a particularbehaviour; helping to form an intention to change a behaviour;setting specific goals; identifying barriers to changing behaviour;planning social support or social change; and providing rewardscontingent on performing the behaviour.
Intervention content and effectsOverall, nine interventions had positive effects, seven resulted inno detectable change and one had an adverse effect. Forsmoking, four studies reported a positive effect,28 39–41 and threereported no effects.34 42 43 For healthy eating, four studiesreported positive effects30 32 38 44 and two reported no effects.33 34
For physical activity, one study reported a positive effect,31 two
no effects32 33 and one an adverse effect.34 Overall, five of thenine positive intervention effects were obtained from RCTsaimed at promoting healthy eating,30 32 38 physical activity31 andsmoking cessation,39 three of which were conducted in theUK.31 38 39 Differing sample sizes do not appear to explaindifferences in effectiveness, as there was no difference betweensample sizes for effective interventions (M = 665, SD = 812) andthose having no effect (M = 613, SD = 443; t(11) = 0.12,p = 0.91).
Effective interventions had a tendency to have fewertechniques on average than ineffective interventions; anindependent t-test demonstrated that this reached marginalstatistical significance (8.22 vs 12.75; t(15) = 21.95, p = 0.07;95% CI for mean difference 29.48 to 0.43) (fig 3). Figure 4compares the frequency with which techniques are included ineffective and ineffective interventions. The most commonlyused techniques were providing information (the first threetechniques in figure 4) and facilitating goal setting (the secondtwo techniques in fig 4). Visual inspection suggests that all thetechniques have potential merit, with none appearing to beoverwhelmingly effective or ineffective.
Effects over timeThe effects of behaviour change interventions may ameliorateover time, such that significant intervention effects are lesslikely to be observed in studies with longer follow-ups. In thisreview, however, there is no evidence of a time-dilution effect,but evidence suggests the reverse. Of the nine positive effectsreported among the included studies, four were obtained instudies with follow-up ,12 months, and five in studies with>12 month follow-ups. Indeed, with only one exception,34
studies with long-term follow-up reported only positiveintervention effects, with non-significant effects thereforeclustered among studies with shorter term follow-up. Ofcourse, results based on indirect comparison must be interpretedwith caution. Nevertheless, it is plausible to speculate that theeffects of interventions targeting low-income groups may takelonger to emerge, and/or have effects that may be more durableor sustainable over time. Resolving these issues may have
Figure 2 Technique type and frequency.
Essay
612 J Epidemiol Community Health 2009;63:610–622. doi:10.1136/jech.2008.078725
on 16 July 2009 jech.bmj.comDownloaded from
profound effects on intervention development, research designand, ultimately, reducing health inequalities among low-incomegroups.
Theoretical base of interventionsTheories were cited in 6 of the 13 included studies, incorporat-ing 10 of the 17 comparisons, four studies drawing on morethan one theory. Five studies cited the stages of change/transtheoretical model, four studies cited social cognitivetheory, and five theories were cited only once: the theory ofreasoned action, the precaution adoption model, the precede–proceed model, behaviour modification principles and organisa-tional theory. There was thus a plethora of theories used inthese few studies. None of the papers reported how the statedtheories were used as a basis for selecting the techniquescomprising the intervention, and none drew conclusions abouttheory from the evaluation data. There were no obviousdifferences in intervention content (ie, the techniques usedwithin the intervention) between those studies that explicitlyused theory and those that did not. There was also no obviousassociation between reported use of theory and whether or notthe intervention was effective.
DISCUSSIONThere is evidence that behaviour change interventions can beeffective in low-income groups. Of the 17 interventionevaluations that targeted low-income groups, we found thatnine were effective, seven were not statistically different andone suggested adverse effects. That nine statistically significantpositive results were obtained purely by chance seems unlikely.However, the small number of studies means that the resultsshould be treated with caution.
The most frequently used intervention techniques wereproviding information (eg, about the consequences of thebehaviour) and prompting people to form intentions and setgoals. No clear patterns between the purported theoretical basis,intervention content and the effect were evident, although thedataset was too small for formal analysis. In addition, the lackof reported detail as to how theory was used in designinginterventions precludes the possibility of explaining a lack ofassociation. It may be that particular theories were not usefulfor intervention development, or that useful theories werepoorly applied.
There are two suggestive findings from this review. The firstis that more focused interventions involving a small set oftechniques may be more effective than interventions combininga large number of different techniques. There may be morevariation in the quality of intervention delivery as the numberof component techniques increases, both within and betweenproviders, increasing the likelihood of inconsistent effects. Thishighlights the need to monitor the fidelity of interventiondelivery,45 46 an important practice that is rarely observed. Moreresearch based on larger datasets is required to clarify theseissues.
The second suggestive finding is that the most commontechniques—providing information, facilitating goal setting andprompting barrier identification—may be helpful for low-income groups. These sets of techniques may be workingadditively, in that providing information about the benefits ofchanging behaviour may increase people’s motivation tochange, while helping people to form specific, realistic goals,identify barriers and draw on social support may help people totranslate motivation into action. This has some parallels with afinding from Coulter and Ellins’ systematic review of patient-focused interventions.47 They found that providing information,on its own, had little effect on people’s knowledge about theirown health. However, combined with professional consultation
Figure 3 Number of techniques used in effective and ineffectiveinterventions.
Figure 4 Number of effective andineffective interventions using eachtechnique.
Essay
J Epidemiol Community Health 2009;63:610–622. doi:10.1136/jech.2008.078725 613
on 16 July 2009 jech.bmj.comDownloaded from
Tabl
e1
Sum
mar
yof
stud
ies
Stu
dyS
tudy
desi
gnC
ount
ryTa
rget
Par
tici
pant
sTh
eore
tica
lba
seof
inte
rven
tion
Con
tent
asde
scri
bed
inst
udy
Res
ults
:ef
fect
s
1.A
ndre
ws
etal
(200
7)(in
terv
entio
nde
scrip
tion
supp
lem
ente
dby
And
rew
set
al20
05)
RC
TU
SA
Sm
okin
gN
:10
3N
one
stat
edC
onte
nt:
Incr
easi
ngse
lf-ef
ficac
yan
dsu
ppor
ting
rela
tions
hips
;fr
eeN
RT,
self-
help
educ
atio
nal
mat
eria
ls,
cultu
rally
spec
ific
Out
com
e:S
mok
ing
stat
us
Incl
usio
n:A
fric
an–A
mer
ican
wom
enD
eliv
ery:
Nur
se-le
dbe
havi
oura
l/em
pow
erm
ent
coun
selli
ngin
grou
p;w
eekl
yin
divi
dual
coun
selli
ngin
pers
onor
byph
one
bytr
aine
dco
mm
unity
heal
thw
orke
rs,
who
wer
elo
cal
Afr
ican
–Am
eric
anw
omen
,ex
-sm
oker
s
Ass
essm
ent:
Sel
f-re
port
and
carb
onm
onox
ide
brea
thre
adin
gs
Indi
cato
r:Pu
blic
hous
ing
deve
lopm
ent
Con
text
:C
omm
unity
Effe
ct:
Posi
tive
Inte
nsity
:W
eekl
y1-
hour
grou
ped
ucat
ion
and
beha
viou
ral
coun
selli
ngfo
r6
wee
ksan
da
boos
ter
coun
selli
ngse
ssio
nat
wee
ks12
and
24.
Con
tact
with
CH
Won
cea
wee
kfo
r24
wee
ks
6-m
onth
abst
inen
cew
as27
.5%
and
5.7%
inin
terv
entio
nan
dco
mpa
rison
grou
ps.
Inte
rven
tion
grou
pm
ore
likel
yto
quit
smok
ing
than
com
paris
onpa
rtic
ipan
ts(O
R=
6.18
,95
%C
I=1.
65to
23.0
)
Follo
w-u
p:12
and
24w
eeks
2.A
usla
nder
etal
(200
2)(in
terv
entio
nde
scrip
tion
supp
lem
ente
dby
Aus
land
eret
al20
00)
RC
TU
SA
Die
tN
:29
4S
tage
sof
chan
geC
onte
nt:
In-c
lass
and
com
mun
ityde
mon
stra
tions
,te
ach
part
icip
ants
how
toas
sess
fat
indi
et,
labe
lre
adin
g,sh
oppi
ngfo
rlo
w-f
atfo
ods,
reci
pem
odifi
catio
n,ea
ting
out,
copi
ngw
ithhi
gh-r
isk
situ
atio
ns;
cultu
rally
spec
ific
Out
com
e:Fa
tin
take
(dai
lyca
lorie
sfr
omfa
tan
dsa
tura
ted
fat)
Incl
usio
n:O
verw
eigh
tA
fric
an–
Am
eric
anw
omen
Del
iver
y:In
divi
dual
and
grou
ppe
erco
unse
lling
bype
ered
ucat
ors
from
Afr
ican
–Am
eric
anco
mm
unity
.A
sses
smen
t,re
view
ofcu
rren
tst
age,
goal
sett
ing
Ass
essm
ent:
Sel
f-re
port
Indi
cato
r:Lo
w-in
com
ear
eaC
onte
xt:
Com
mun
ityEf
fect
:Po
sitiv
e;pe
rcen
tage
ofca
lorie
sfr
omfa
tfo
rtr
eatm
ent
grou
pw
asre
duce
dfr
om35
.9%
pre-
test
to32
.3%
at6-
mon
thfo
llow
-up
com
pare
dw
ith36
%an
d34
.5%
,re
spec
tivel
y,fo
rco
ntro
lsu
bjec
ts.
No
SD
.D
iffer
ence
betw
een
cont
rol
and
trea
tmen
t:at
base
line,
t=0.
03n.
s.;
3m
onth
spo
st,
t=4.
01,
p,0.
001;
6m
onth
sfo
llow
-up
,t=
2.50
,p,
0.05
.C
alor
ies
from
satu
rate
dfa
ttr
eatm
ent
grou
psi
gnifi
cant
lyre
duce
dfr
ompr
e12
.4%
topo
st10
.8%
and
follo
w-u
p10
.9%
.C
ontr
olsc
ored
12.4
%,
12.3
%,
12.0
%re
spec
tivel
y.D
iffer
ence
betw
een
cont
rol
and
trea
tmen
t:at
base
line,
t=0.
00n.
s.;
3m
onth
spo
st,
t=4.
39,
p,0.
001;
6m
onth
sfo
llow
-up
t=3.
00p,
0.01
Inte
nsity
:12
sess
ions
(six
grou
p,si
xin
divi
dual
),on
epe
rw
eek
for
3m
onth
s—se
ssio
nle
ngth
not
repo
rted
Follo
w-u
p:6
mon
ths
Con
tinue
d
Essay
614 J Epidemiol Community Health 2009;63:610–622. doi:10.1136/jech.2008.078725
on 16 July 2009 jech.bmj.comDownloaded from
Tabl
e1
Con
tinue
d
Stu
dyS
tudy
desi
gnC
ount
ryTa
rget
Par
tici
pant
sTh
eore
tica
lba
seof
inte
rven
tion
Con
tent
asde
scri
bed
inst
udy
Res
ults
:ef
fect
s
3.B
emel
man
set
al(2
000)
Con
trol
led
com
paris
onst
udy
Net
herla
nds
Die
tN
:26
6S
tage
sof
chan
geC
onte
nt:
Info
rmat
ion
abou
the
alth
ypr
oduc
tsan
dM
edite
rran
ean
diet
,ex
plan
atio
nof
read
ing
food
labe
ls,
vide
oab
out
shop
ping
for
food
and
prod
uct
choi
ce,
loca
lre
cipe
book
,pe
rson
alle
tter
tailo
red
tost
age
ofch
ange
prov
idin
gta
ilore
din
form
atio
n
Out
com
e:C
onsu
mpt
ion
offr
uit,
vege
tabl
es,
red
mea
tan
din
take
ofto
tal
and
satu
rate
dfa
t
Incl
usio
n:H
yper
chol
este
rola
emic
adul
tsw
ithat
leas
ttw
oad
ditio
nal
CV
Dris
kfa
ctor
s
Del
iver
y:S
mal
lgr
oup
sess
ions
(n=
10)
with
part
ners
atte
ndin
g,an
dbo
okle
tsw
ithpr
ogra
mm
e-re
leva
ntin
form
atio
n
Ass
essm
ent:
Sel
f-re
port
Indi
cato
r:S
ocio
econ
omic
ally
depr
ived
area
Con
text
:C
omm
unity
Effe
ct:
Frui
t=
posi
tive
outc
ome
for
men
inte
rven
tion
mea
n29
7g/
day
(SD
170)
,w
omen
inte
rven
tion
342
(180
),m
enco
ntro
l22
1(1
63),
wom
enco
ntro
l28
3(1
75)
Inte
nsity
:Th
ree
2-ho
urse
ssio
ns+4
diet
ary
educ
atio
nbo
okle
tsTo
tal
fat
=po
sitiv
e—re
duce
inta
kew
ithm
ean
net
diff
eren
ceof
1.8%
Follo
w-u
p:52
wee
ksS
atur
ated
fat
=po
sitiv
ene
tdi
ffer
ence
1.1%
4.Em
mon
set
al(2
005)
RC
TU
SA
Die
t,ex
erci
seN
:19
54S
ocia
lco
gniti
veth
eory
,st
ages
ofch
ange
,pr
ecau
tion
adop
tion
and
orga
nisa
tiona
lth
eorie
s
Con
tent
:Pr
ovis
ion
ofa
tailo
red
pres
crip
tion
for
canc
erpr
even
tion
indi
catin
gsp
ecifi
cris
kfa
ctor
s,in
divi
dual
coun
selli
ngus
ing
mot
ivat
iona
lin
terv
iew
ing
Out
com
e:D
iet
(fru
itan
dve
geta
bles
per
day,
red
mea
tco
nsum
ptio
n)an
dex
erci
se(w
eekl
ym
inut
esof
phys
ical
activ
ity)
Ass
essm
ent:
Sel
f-re
port
Incl
usio
n:N
on-r
elat
edm
edic
alco
nsul
tatio
nD
eliv
ery:
Tele
phon
eco
unse
lling
and
writ
ten
mat
eria
lEf
fect
:D
iet
=po
sitiv
e:A
djus
ted
%of
part
icip
ants
with
each
heal
thbe
havi
our
atba
selin
ean
dfo
llow
-up;
Ser
ving
sfr
uit/
veg
>5/
dC
ontr
ol2
3.8%
Inte
rven
tion
=3.
3%p
=0.
005
Indi
cato
r:Lo
w-in
com
ear
eaC
onte
xt:
Prim
ary
care
Exer
cise
=no
n-si
gnifi
cant
Inte
nsity
:O
ne20
-min
ute
indi
vidu
alco
unse
lling
sess
ion,
four
10-m
inut
efo
llow
-up
tele
phon
eco
unse
lling
sess
ions
,si
xse
tsof
tailo
red
writ
ten
mat
eria
l+
ongo
ing
activ
ities
Follo
w-u
p:8
mon
ths
5.Fi
sher
etal
(199
8)Q
uasi
-ex
perim
enta
lU
SA
Sm
okin
gN
:22
19–7
neig
hbou
rhoo
ds(t
wo
citie
s)m
atch
edfo
ret
hnic
ity,
inco
me
and
educ
atio
n
Non
est
ated
Con
tent
:U
ncle
arO
utco
me:
Sm
okin
gpr
eval
ence
Incl
usio
n:R
esid
ents
(80%
Afr
ican
–A
mer
ican
)co
ntac
tabl
eby
tele
phon
e
Del
iver
y:M
ultip
le:
incl
uded
smok
ing
cess
atio
ncl
asse
s,bi
llboa
rds,
door
-to-
door
cam
paig
n,go
spel
fest
with
anti-
smok
ing
song
s,he
alth
fairs
,se
lf-he
lpbr
ochu
res
Ass
essm
ent:
Sel
f-re
port
Indi
cato
r:Lo
w-in
com
ear
eaC
onte
xt:
Com
mun
ityEf
fect
:Po
sitiv
e:S
mok
ing
prev
alen
cede
clin
ed7%
p=
0.02
8in
St
Loui
spr
ogra
mm
ere
ceiv
ing
sam
ple
and
only
1%p
=0.
641
inKa
nsas
City
com
paris
ongr
oup
betw
een
1990
and
1992
.R
elia
bilit
ysu
gges
ted
byre
duce
dpr
eval
ence
inth
ree
St
Loui
sne
ighb
ourh
oods
Con
tinue
d
Essay
J Epidemiol Community Health 2009;63:610–622. doi:10.1136/jech.2008.078725 615
on 16 July 2009 jech.bmj.comDownloaded from
Tabl
e1
Con
tinue
d
Stu
dyS
tudy
desi
gnC
ount
ryTa
rget
Par
tici
pant
sTh
eore
tica
lba
seof
inte
rven
tion
Con
tent
asde
scri
bed
inst
udy
Res
ults
:ef
fect
s
Inte
nsity
:U
ncle
ar,
but
prog
ram
me
ran
for
24m
onth
s
Follo
w-u
p:2
year
sfr
omst
art
ofpr
ogra
mm
e
6.H
ahn
etal
(200
4)Q
uasi
-exp
erim
enta
lU
SA
Sm
okin
gN
:53
8(in
terv
entio
nn
=24
8;co
ntro
ln
=29
0)N
otst
ated
Con
tent
:U
ncle
arO
utco
me:
Toba
cco
use
Incl
usio
n:N
oon
eex
clud
ed,
volu
ntee
rsD
eliv
ery:
Mul
tiple
—Q
uit
and
win
supp
orte
dby
prov
ider
advi
ce,
mai
led
post
card
s,on
line
and
1:1
tele
phon
equ
itas
sist
ance
,med
iaca
mpa
ign
(rad
io,
tele
visi
onad
vert
isem
ents
,bi
llboa
rdan
dne
wsp
aper
feat
ures
,pr
omot
iona
lfli
ers)
,gr
oup
smok
ing
cess
atio
ncl
asse
s,co
mm
unity
quit
date
and
cash
priz
elo
tter
y
Ass
essm
ent:
Sel
f-re
port
ed—
not
havi
ngus
edan
yfo
rmof
toba
cco
with
inpa
st7
days
;ur
ine
sam
ple
Indi
cato
r:H
adto
be18
year
sor
olde
ran
dea
rnle
ssth
an$2
5,00
0pe
rye
ar
Con
text
:C
omm
unity
Effe
ct:
Posi
tive:
Con
firm
edqu
itra
tes
show
edth
at,a
t3
mon
ths,
11%
had
quit
com
pare
dw
ith0.
7%an
d,at
12m
onth
s,8%
had
quit
com
pare
dw
ith0.
7%
Inte
nsity
:U
ncle
ar
Follo
w-u
p:12
mon
ths
7.Lo
wth
eret
al(2
002)
(inte
rven
tion
desc
riptio
nsu
pple
men
ted
by3
6)
RC
TU
Kph
ysic
alac
tivity
N:
370
Not
clea
rC
onte
nt:
Free
vouc
hers
tolo
cal
exer
cise
faci
litie
s.Ex
erci
seco
nsul
tatio
nex
perim
enta
lgr
oup
rece
ived
a1:
1ex
erci
seco
nsul
tatio
nin
terv
iew
whi
chdi
scus
sed
how
tobe
com
em
ore
phys
ical
lyac
tive,
adva
ntag
esan
ddi
sadv
anta
ges
ofch
ange
,ba
rrie
rs,
soci
alsu
ppor
t,go
alse
ttin
g,re
laps
epr
even
tion
Out
com
e:R
egul
arph
ysic
alac
tivity
Incl
usio
n:re
side
nts
oftw
oS
cott
ish
hous
ing
esta
tes
Del
iver
y:Fi
tnes
sas
sess
men
tex
perim
enta
lgr
oup
rece
ived
stan
dard
com
pute
rised
phys
ical
fitne
ssas
sess
men
tan
dth
enof
fere
dan
exer
cise
prog
ram
me
gear
edto
pers
onal
capa
bilit
ies.
Exer
cise
cons
ulta
tion
got
1:1
inte
rvie
w.
Con
trol
grou
pgo
the
ight
and
body
mas
sm
easu
rem
ent
and
info
rmat
ion
onph
ysic
alac
tivity
.
Con
text
:co
mm
unity
Ass
essm
ent:
Sel
f-re
port
Indi
cato
r:Lo
w-in
com
ear
ea;p
eopl
eno
tre
gula
rlyac
tive
Inte
nsity
:30
-min
ute
cons
ulta
tion
Effe
ct:N
odi
ffer
ence
inph
ysic
alac
tivity
betw
een
two
grou
ps.
Phys
ical
activ
ityin
crea
sed
from
base
line
to4
wee
ksan
dm
aint
aine
dat
3an
d6
mon
ths
for
both
grou
ps,
but
decl
ined
by12
mon
ths
follo
w-u
p.O
nly
exer
cise
cons
ulta
tion
expe
rimen
tal
grou
pre
port
edsi
gnifi
cant
lym
ore
activ
ityaf
ter
1ye
ar.
Dro
pout
rate
for
fitne
ssas
sess
men
tgr
oup
was
high
erth
anfo
rex
erci
seco
nsul
tatio
ngr
oup
Follo
w-u
p:4,
12,
24,
52w
eeks
8.O
’Lou
ghlin
etal
(199
9)[in
terv
entio
nde
scrip
tion
supp
lem
ente
dby
37)
Qua
si-
expe
rimen
tal
Can
ada
Sm
okin
g,ph
ysic
alac
tivity
,di
etN
:11
95ne
ighb
ourh
oods
(tw
om
atch
edar
eas)
Soc
ial
lear
ning
theo
ry,
the
reas
oned
actio
nm
odel
,an
dth
epr
eced
e–pr
ocee
dm
odel
Del
iver
y:M
ultip
le:
smok
ing
cess
atio
nw
orks
hops
,he
art
heal
thre
cipe
cont
ests
,he
art
heal
thed
ucat
ion
wor
ksho
ps,
men
ula
belli
ng,
dire
ctm
ail
and
vide
oed
ucat
iona
lm
ater
ials
Out
com
e:S
mok
ing,
diet
(fat
inta
ke),
phys
ical
activ
ity
Con
tinue
d
Essay
616 J Epidemiol Community Health 2009;63:610–622. doi:10.1136/jech.2008.078725
on 16 July 2009 jech.bmj.comDownloaded from
Tabl
e1
Con
tinue
d
Stu
dyS
tudy
desi
gnC
ount
ryTa
rget
Par
tici
pant
sTh
eore
tica
lba
seof
inte
rven
tion
Con
tent
asde
scri
bed
inst
udy
Res
ults
:ef
fect
s
Incl
usio
n:R
esid
ents
cont
acta
ble
byte
leph
one
Con
text
:C
omm
unity
Ass
essm
ent:
Sel
f-re
port
Indi
cato
r:Lo
w-in
com
ear
eaw
ithhi
ghC
VD
mor
talit
yra
teIn
tens
ity:
Unc
lear
but
prog
ram
me
ran
for
48m
onth
sEf
fect
:
(a)
Sm
okin
g:N
on-s
igni
fican
tFo
llow
-up:
3ye
ars
and
5ye
ars
from
the
star
tof
the
prog
ram
me
(b)
Die
t(f
atin
take
):N
on-s
igni
fican
t(c
)Ph
ysic
alac
tivity
:N
egat
ive
9.R
osam
ond
etal
(200
0)N
on-
rand
omis
edin
terv
entio
ntr
ial
US
AD
iet,
phys
ical
activ
ityN
:71
2(5
79en
hanc
edin
terv
entio
ngr
oup,
133
min
imum
inte
rven
tion
grou
p)
Soc
ial
cogn
itive
theo
ry,
the
tran
sthe
oret
ical
mod
elan
dbe
havi
our
mod
ifica
tion
prin
cipl
es
Con
tent
:Pa
tient
asse
ssm
ent
tose
tgo
als,
coun
selli
ngto
impr
ove
patie
ntse
lf-ef
ficac
y,ed
ucat
iona
lm
ater
ial
toid
entif
yan
dre
info
rce
posi
tive
beha
viou
rs,
tailo
red
tipsh
eets
,pa
tient
self-
asse
ssm
ent,
indi
vidu
alco
unse
lling
,br
eaki
nggo
als
into
smal
l,ac
hiev
able
step
s
Out
com
e:D
iet
(fat
inta
ke);
phys
ical
activ
ity(f
requ
ency
,du
ratio
n,in
tens
ity)
Incl
usio
n:al
lw
omen
50ye
ars
ofag
eor
olde
r,in
com
e,
200%
ofpo
vert
yle
vel,
unin
sure
dor
unde
rinsu
red
resi
dent
inon
eof
31co
untie
s(1
7m
inim
umin
terv
entio
n,14
enha
nced
inte
rven
tion)
foun
dto
beat
elev
ated
risk
for
CV
D.
Del
iver
y:M
inim
umin
terv
entio
n(M
I):
heal
thde
part
men
tus
ual
coun
selli
ngsy
stem
and
educ
atio
nm
ater
ials
Ass
essm
ent:
Sel
f-re
port
Indi
cato
r:Lo
win
com
eEn
hanc
edin
terv
entio
n(E
I):
Hea
lthde
part
men
tst
aff
prov
ided
ast
ruct
ured
asse
ssm
ent
ofdi
etan
dph
ysic
alac
tivity
follo
wed
byst
ruct
ured
coun
selli
ngfo
cuse
don
beha
viou
rsin
grea
test
need
ofch
ange
Effe
ct:
Con
text
:Pr
imar
yca
re(a
)D
iet:
Posi
tive:
redu
ced
chol
este
rol,
bloo
dpr
essu
re,
smok
ing
inbo
thgr
oups
alth
ough
not
stat
istic
ally
sign
ifica
nt
Inte
nsity
:M
I—us
ualh
ealth
depa
rtm
ent
follo
w-
up:
varia
ble
and
not
spec
ified
(b)
Phys
ical
activ
ity:
Posi
tive—
5.9%
repo
rtre
gula
rph
ysic
alac
tivity
inEI
grou
pan
d9.
3%in
MI
grou
pbu
tdi
ffer
ence
betw
een
grou
psw
asno
tst
atis
tical
lysi
gnifi
cant
EI:
Thre
ein
terv
entio
nvi
sits
Follo
w-u
p:7
mon
ths
10.
Sol
omon
etal
(200
0)R
CT
US
AS
mok
ing
N:
214
Not
stat
edC
onte
nt:
Unc
lear
Out
com
e:S
mok
ing
prev
alen
ce
Incl
usio
n:D
eliv
ery:
Free
NR
T+
tele
phon
eco
unse
lling
Ass
essm
ent:
Sel
f-re
port
and
carb
onm
onox
ide
read
ings
Wom
enag
ed18
–50
year
s,w
hosm
oked
.fo
urci
gare
ttes
per
day
and
had
aho
me
tele
phon
e
Con
text
:H
ome
Effe
ct:
Non
-sig
nific
ant.
At
3m
onth
sfo
llow
-up,
42%
ofw
omen
inpa
tch
plus
proa
ctiv
eco
unse
lling
wer
eab
stin
ent
com
pare
dw
ith28
%of
patc
hon
ly.
No
diff
eren
ces
at6
mon
ths
follo
w-u
p.Ex
perim
enta
lsu
bjec
tsre
port
grea
ter
conf
iden
ceto
quit
Con
tinue
d
Essay
J Epidemiol Community Health 2009;63:610–622. doi:10.1136/jech.2008.078725 617
on 16 July 2009 jech.bmj.comDownloaded from
Tabl
e1
Con
tinue
d
Stu
dyS
tudy
desi
gnC
ount
ryTa
rget
Par
tici
pant
sTh
eore
tica
lba
seof
inte
rven
tion
Con
tent
asde
scri
bed
inst
udy
Res
ults
:ef
fect
s
Indi
cato
r:Lo
win
com
eIn
tens
ity:
8–10
wee
ksof
nico
tine
patc
hes
+on
eph
one
call
befo
requ
itda
y,on
quit
day,
one
phon
eca
ll4
days
afte
rqu
itda
yan
da
wee
kly
orbi
wee
kly
phon
eca
llfo
rup
to3
mon
ths
Follo
w-u
p:6
mon
ths
11.
Sol
omon
etal
(200
5)R
CT
US
AS
mok
ing
N:
330
Not
stat
edC
onte
nt:
free
NR
T+
proa
ctiv
ete
leph
one
peer
supp
ort,
the
supp
ort
pers
onfo
llow
eda
sem
i-st
ruct
ured
prot
ocol
topr
ovid
een
cour
agem
ent,
guid
ance
and
rein
forc
emen
tfo
rqu
ittin
gsm
okin
gan
dto
assi
stin
prob
lem
solv
ing
high
risk
for
smok
ing
situ
atio
ns
Out
com
e:S
mok
ing
cess
atio
n
Incl
usio
n:W
omen
aged
18–
50ye
ars,
smok
ing
.fo
urci
gare
ttes
ada
y,w
ithho
me
tele
phon
e,in
tend
ing
toqu
itin
the
next
2w
eeks
Del
iver
y:Te
leph
one
inte
rvie
wan
dfr
eeni
cotin
epa
tche
sth
roug
hth
epo
stA
sses
smen
t:S
elf-
repo
rted
abst
inen
ce
Indi
cato
r:C
urre
ntly
rece
ivin
gM
edic
aid
orV
erm
ont
Hea
lthA
ssis
tanc
ePl
an(V
HA
P)he
alth
care
cove
rage
for
low
-inco
me
Ver
mon
tre
side
nts
Con
text
:H
ome
Effe
ct:
Non
-sig
nific
ant—
At
3m
onth
s,te
leph
one
supp
ort
had
asi
gnifi
cant
effe
ctw
ith43
%of
expe
rimen
talv
s26
%of
cont
rol
subj
ects
repo
rtin
g30
-day
abst
inen
ce.
Diff
eren
cew
asno
long
ersi
gnifi
cant
at6
mon
ths
Inte
nsity
:U
pto
4m
onth
sof
proa
ctiv
esu
ppor
tby
tele
phon
e.Th
ree
phon
eca
llsth
ensu
bseq
uent
calls
occu
rrin
gw
eekl
yan
dta
perin
gto
biw
eekl
yfo
rup
to4
mon
ths.
Up
to10
wee
kssu
pply
ofni
cotin
epa
tche
s
Follo
w-u
p:6
mon
ths
12.
Ste
ptoe
etal
(200
3)R
CT
UK
Die
tN
:27
1S
ocia
llea
rnin
gth
eory
and
stag
esof
chan
geC
onte
nt:
Beh
avio
ural
coun
selli
ngO
utco
me:
Die
t(f
ruit
and
vege
tabl
eco
nsum
ptio
n)
Incl
usio
n:Pr
imar
yhe
alth
cent
repa
tient
s,ag
ed18
–70
year
s,re
side
ntin
low
-inco
me
area
,w
ithan
inco
me
less
than
£400
per
wee
k
Del
iver
y:A
rese
arch
nurs
epr
ovid
eda
base
line
asse
ssm
ent,
two
indi
vidu
alco
unse
lling
sess
ions
Ass
essm
ent:
Sel
f-re
port
and
biom
arke
rsof
frui
tan
dve
geta
ble
inta
ke(u
rine
test
)
Indi
cato
r:Lo
w-in
com
ear
eaC
onte
xt:
Prim
ary
care
Effe
ct:
Posi
tive.
Con
sum
ptio
nof
frui
tan
dve
geta
bles
incr
ease
dby
1.5
and
0.9
port
ions
per
day
inth
ebe
havi
oura
lan
dnu
triti
ongr
oups
.Pr
opor
tion
ofpa
rtic
ipan
tsea
ting
five
orm
ore
port
ions
per
day
incr
ease
dby
42%
and
27%
inth
etw
ogr
oups
Inte
nsity
:O
neba
selin
eas
sess
men
t,on
e15
-m
inut
eco
unse
lling
sess
ion,
ase
cond
15-
min
ute
coun
selli
ngse
ssio
n2
wee
ksla
ter
Follo
w-u
p:12
mon
ths
Con
tinue
d
Essay
618 J Epidemiol Community Health 2009;63:610–622. doi:10.1136/jech.2008.078725
on 16 July 2009 jech.bmj.comDownloaded from
or advice, it could improve knowledge and recall, especiallywhere the information was personalised. Disadvantaged popu-lations benefited more than other groups, possibly because theirknowledge base was smaller, and so they had more to gain fromhealth information. It is also consistent with a meta-analysis ofinterventions to increase HIV preventive behaviours.16 Providinginformation changed behaviour only when accompanied byactive, behavioural strategies such as teaching self-managementtechniques.
Goal setting is a key behaviour change technique in evidence-based theories of behaviour change, most notably socialcognitive theory48 and self-regulation (control) theory.49
Setting goals that are realistic and achievable helps people tofeel more confident about being able to change their behaviour.Setting goals may also help people to be more aware of theircurrent behaviour and to take steps when they notice theirbehaviour is falling short of their targets. Breaking down large,long-term goals into smaller, short-term goals allows people tobuild on small successes, leading to greater feelings of control or‘‘mastery’’. This may be especially important for those indisadvantaged situations, who often experience a lack of controland therefore feel powerless to bring about change. Emergingresearch findings suggest that adults with a low income or ahigh school education or less score poorly on the ‘‘patientactivation measure’’. This measures an individual’s confidence,knowledge and skills to take action to improve their health andstay the course even under stress.50 Goal setting is a relativelysimple technique that can be successfully taught to a wide rangeof people varying in educational and social background, butdisadvantaged groups may have more to gain, if their confidenceand skill base is lower.51–54
Our scoping review is, by definition, not exhaustive. Giventhat only three of the 13 interventions were conducted in theUK and nine were from North America, caution needs to beexercised about the generalisability of the findings. In addition,the scope of the review did not extend to consider studies thatdirectly compared the effectiveness of an intervention in low-income groups versus more affluent groups. It is therefore notclear whether interventions to change these behaviours aredifferentially effective across socioeconomic groups. Tworeviews have used a similar taxonomy of techniques forinvestigating intervention techniques in populations notselected by socioeconomic status and were able to use meta-regression in much larger datasets.12 55 They identified self-monitoring of behaviour as the most effective technique.However, within-study comparisons are needed to answer thequestion as to whether different techniques, or different modesof delivery, tailored so as to be more relevant or attractive, areneeded to promote health among lower income groups.
However, this review shows that there is a widespreadpaucity of evidence about the effectiveness and cost-effective-ness of changing health behaviours in disadvantaged groups. Tobuild evidence about ‘‘what works for whom’’, it is essentialthat the same intervention be compared across different groups,and that different interventions be compared in the samegroups. As this review demonstrates, such work is in its infancy.
For those with responsibility for commissioning or conduct-ing systematic reviews in this or related areas, there is a need toconsider carefully methods for maximising the inclusion ofrelevant evidence. First, reviews may need to include otherstudy designs that adequately balance methodological reliabilityand contextual relevance. Second, future reviews may need toconsider evidence obtained from a wider range of individuals.For example, our review focused on evidence that reportedTa
ble
1C
ontin
ued
Stu
dyS
tudy
desi
gnC
ount
ryTa
rget
Par
tici
pant
sTh
eore
tica
lba
seof
inte
rven
tion
Con
tent
asde
scri
bed
inst
udy
Res
ults
:ef
fect
s
13.
Syk
esan
dM
arks
(200
1)R
CT
UK
Sm
okin
gN
:26
0U
ncle
arC
onte
nt:
Mul
tiple
Out
com
e:S
mok
ing
prev
alen
ce
Incl
usio
n:S
mok
ers
volu
ntee
ring
toat
tend
asm
okin
gce
ssat
ion
tria
lD
eliv
ery:
Cho
ice
ofps
ycho
logi
cal
met
hods
+op
tiona
lN
RT
Ass
essm
ent:
Sel
f-re
port
and
carb
onm
onox
ide
read
ings
Indi
cato
r:Lo
w-in
com
ear
ea30
CB
Tan
dot
her
met
hods
inse
lf-he
lppa
ckag
eco
nsis
ting
ofha
ndbo
ok,
redu
ctio
nca
rds,
prog
ress
char
t
Effe
ct:
Posi
tive
at6
mon
ths
follo
w-u
p.17
.2%
ofth
ose
rece
ivin
gth
erap
yw
ere
abst
inen
tan
d11
.5%
had
redu
ced
cons
umpt
ion
bya
quar
ter
ofpr
e-tr
eatm
ent
leve
lco
mpa
red
with
5.6%
ofco
ntro
lgr
oup
who
wer
eab
stin
ent
and
none
ofw
hom
had
redu
ced
thei
rco
nsum
ptio
n
Con
text
:C
omm
unity
Inte
nsity
:R
educ
tion:
initi
al7–
10da
ype
riod.
Rel
apse
prev
entio
n:un
clea
rbu
tto
tals
3m
onth
s
Follo
w-u
p:6
mon
ths
NR
T,ni
cotin
ere
plac
emen
tth
erap
y;C
HW
,co
mm
unity
heal
thw
orke
r;C
BT,
cogn
itive
beha
viou
ralt
hera
py;
RC
T,ra
ndom
ised
cont
rolle
dtr
ial;
CV
D,
card
iova
scul
ardi
seas
e;O
R,
odds
ratio
;C
I,co
nfid
ence
inte
rval
;t,
t-st
atis
ticof
mea
ndi
ffer
ence
s;S
D,
stan
dard
devi
atio
n;p,
prob
abili
tyle
vel.
Essay
J Epidemiol Community Health 2009;63:610–622. doi:10.1136/jech.2008.078725 619
on 16 July 2009 jech.bmj.comDownloaded from
outcomes in adults, excluding evidence that targeted low SESadults but reported outcomes only in their children. Third,future reviews should consider the implications of usingdifferent indicators of social disadvantage. For example, in ourreview, the population of interest was defined as low income,but that meant the review excluded ‘‘culturally sensitive/targeted’’ interventions that were not explicitly indexed underterms related to SES.
A dedicated stream of research funding for research intointerventions targeting health behaviour change among low SESgroups would thus seem to be timely and warranted. For suchinvestment to maximise its potential to improve populationhealth, the study of interventions to change behaviour should beinformed by methods for analysing interventions by theircomponent techniques and underlying theories of behaviourchange. This will facilitate building evidence about, not just‘‘what works’’ but how interventions work, evidence that iscrucial for the future development of more effective interventions.
Funding: This review was funded by the King’s Fund.
Competing interests: None.
REFERENCES1. World Cancer Research Fund, American Institute of Cancer Research. Food,
nutrition, physical activity and the prevention of cancer: a global perspective.Washington DC: Institute of Cancer Research, 2007.
2. World Health Organization. The world health report 2002. Reducing risks to health,promoting healthy life. Geneva: World Health Organization, 2002.
3. Kawachi I, Kennedy BP, Wilkinson RG, eds. The society and population healthreader. Vol. 1. Income inequality and health. New York: New Press, 1999
4. Marmot M. Status syndrome: how your social standing directly affects your healthand life expectancy. London: Bloomsbury, 2004.
5. Wilkinson RG. Unhealthy societies: the afflictions of inequality. London: Routledge,1996.
6. Wilkinson RG, Marmot M, eds. Social determinants of health. Oxford: OxfordUniversity Press, 1999.
7. Office of National Statistics. Prevalence of cigarette smoking by sex and socio-economic classification based on the current or last job of the household referenceperson. http://www.statistics.gov.uk/hub/health-social-care/index.html. 2005(accessed 26 May 2009).
8. NHS Health and Social Care Information Centre. Health survey for England:health of ethnic minorities. http://www.ic.nhs.uk/statistics-and-data-collections/health-and-lifestyles-related-surveys/health-survey-for-england/health-survey-for-england-2004-health-of-ethnic-minorities-headline-tables. 2004.
9. Office of National Statistics. Census 2001. http://www.statistics.gov.uk/census/default.asp. 2001 (accessed 26 Feb 2008).
10. Office of National Statistics. Labour Force Survey, Table 3.6 D7665. http://www.statistics.gov.uk/STATBASE/Product.asp?vlnk=14248. 2003 (accessed 26 Feb 2008).
11. Hillsdon M, Foster C, Caill N, et al. The effectiveness of public health interventions forincreasing physical activity among adults: a review of reviews, 2nd edn. London:Health Development Agency, 2005.
12. Michie S, Abraham C, Whittington C, et al. Effective techniques in healthy eatingand physical activity interventions: a meta-regression. Health Psychol in press.
13. Naidoo B, Warm D, Quigley R, et al. Smoking and public health: a review of reviewsof interventions to increase smoking cessation, reduce smoking initiation and preventfurther uptake of smoking. London: Health Development Agency, 2004.
14. Blaxter M. Evidence for the effect on inequalities in health of interventions designedto change behaviour. http://www.nice.org.uk/nicemedia/pdf/EvidencefortheeffectonInequalitiesdesignedtochangebehavior.pdf. 2007 (accessed 19Feb 2008).
15. National Institute for Health and Clinical Excellence. Behaviour change atpopulation, community and individual levels. Public health programme guidance 6.London: National Institute of Health and Clinical Excellence, 2007.
16. Albarracin D, Gillette JC, Earl AN, et al. A test of major assumptions aboutbehaviour change: a comprehensive look at the effects of passive and active HIV-prevention interventions since the beginning of the epidemic. Psychol Bull2005;131:856–97.
17. Kidd KE, Altman DG. Adherence in social context. Controlled Clin Trials2000;21:694–6.
18. Ferguson J, Bauld L, Chesterman J, et al. The English smoking treatment services:one-year outcomes. Addiction 2005;100:59–69.
19. Bell K, McCullough L, Greaves L. The effectiveness of National Health Serviceintensive treatments for smoking cessation in England: a systematic review. London:National Institute for Clinical and Health Excellence, 2006.
20. Bernal G, Sharron-del-Rio MR. Are empirically supported treatments valid for ethnicminorities? Toward an alternative approach for treatment research. Cult DiversityEthnic Minority Psychol 2001;7:328–42.
21. Yancey AK, Ortega AN, Kumanyika SK. Effective recruitment and retention ofminority research participants. Annu Rev Public Health 2006;27:1–28.
22. Mokdad AH, Marks JS, Stroup DF, et al. Actual causes of death in the UnitedStates, 2000 (vol. 291, pp. 1238, 2004). J Am Med Assoc 2004;293:293–4.
23. Michie S. Designing and implementing ‘‘behaviour change’’ interventions to improvepopulation health. J Health Serv Res Policy 2008;13:64–9.
24. Michie S, Johnston M, Francis J, et al. From theory to intervention: mappingtheoretically derived behavioural determinants to behaviour change techniques. ApplPsychol Int Rev 2008;57:660–80.
25. Michie S, Abraham C. Identifying techniques that promote health behaviour change:evidence based or evidence inspired? Psychol Health 2004;19:29–49.
26. Campbell M, Fitzpatrick R, Haines A, et al. Framework for design and evaluation ofcomplex interventions to improve health. BMJ 2000;321:694–6.
27. Andrews JO, Bentley DNP, Crawford CHW, et al. Using community-basedparticipatory research to develop a culturally sensitive smoking cessation interventionwith public housing neighbours. Ethnicity Disease 2007;17:331–7.
28. Andrews JO, Felton G, Wewers ME, et al. Sister to sister: a pilot study to assistAfrican American women in subsidized housing to quit smoking. SouthernOnline J Nursing Res 2005;6:23–31.
29. Auslander W, Haire-Joshu D, Houston C, et al. The short term impact of a healthpromotion program for low-income African American women. Res Social Work Pract2000;10:78–97.
30. Auslander W, Haire-Joshu D, Houston C, et al. A controlled evaluation of stagingdietary patterns to reduce the risk of diabetes in African-American women. DiabetesCare 2002;25:809–14.
31. Lowther M, Mutrie N, Scott EM. Promoting physical activity in a socially andeconomically deprived community: a 12-month randomized control trial of fitnessassessment and exercise consultation. J Sport Sci 2002;20:577–88.
32. Emmons KM, Stoddard AM, Fletcher R, et al. Cancer prevention among workingclass, multiethnic adults: results of the healthy directions-health centers study.Am J Public Health 2005;95:1200–5.
33. Rosamond WD, Ammerman AS, Holliday JL, et al. Cardiovascular disease risk factorintervention in low-income women: the North Carolina WISEWOMAN project. PrevMed 2000;31:370–9.
34. O’Loughlin J, Paradis G, Gray-Donald K, et al. The impact of a community-basedheart disease prevention program in a low-income, inner-city neighborhood.Am J Public Health 1999;89:1819–26.
35. Abraham C, Michie S. A taxonomy of behavior change techniques used ininterventions. Health Psychol 2008;27:379–87.
36. Loughlan C, Mutrie N. Conducting an exercise consultation: guidelines for healthprofessionals. J Inst Health Educ 1995;33:78–82.
37. Paradis G, O’Loughlin J, Elliott M, et al. A heart health promotion programme in alow income, low education neighbourhood in Montreal, Canada: theoretical modeland early field experience. J Epidemiol Community Health 1995;49:503–12.
38. Steptoe AL, Perkins-Porras EA, McKay C, et al. Behavioural counselling to increaseconsumption of fruit and vegetables in low income adults: randomised trial. BMJ2003;326:855–8.
39. Sykes CM, Marks DF. Effectiveness of a cognitive behaviour therapy self-helpprogramme for smokers in London, UK. Health Promotion Int 2001;16:255–60.
40. Fisher EB, Auslander WF, Munro JF, et al. Neighbors for a smoke free north side:evaluation of a community organization approach to promoting smoking cessationamong African Americans. Am J Public Health 1998;88:1658–63.
41. Hahn EJ, Rayens MK, Chirila C, et al. Effectiveness of a quit and win contest with alow-income population. Prev Med 2004;39:543–50.
42. Solomon LJ, Scharoun GM, Flynn BS, et al. Free nicotine patches plus proactivetelephone peer support to help low-income women stop smoking. Prev Med2000;31:68–74.
43. Solomon LJ, Marcy TW, Howe KD, et al. Does extended proactive telephonesupport increase smoking cessation among low-income women using nicotinepatches? Prev Med 2005;40:306–13.
What this study adds
c There are few well-conducted evaluations of interventions tochange health behaviours targeted at low-income groups.
c An innovative method of specifying intervention content usinga reliable taxonomy of techniques.
c The most frequently used techniques included goal setting,planning social support/change and providing rewards.
c Such interventions can be effective for low-income groups.c There was no difference in intervention content or outcome
between studies that said they used theory and those that didnot.
Essay
620 J Epidemiol Community Health 2009;63:610–622. doi:10.1136/jech.2008.078725
on 16 July 2009 jech.bmj.comDownloaded from
44. Bemelmans WJ, Broer J, Vries JH, et al. Impact of Mediterranean diet educationversus posted leaflet on dietary habits and serum cholesterol in high risk populationfor cardiovascular disease. Public Health Nutr 2000;3:273–83.
45. Bellg AJ, Borrelli B, Resnick B, et al. Enhancing treatment fidelity in health behaviorchange studies: best practice and recommendations from the Behavior ChangeConsortium. Health Psychol 2004;23:443–51.
46. Hardeman W, Michie S, Fanshawe T, et al. Fidelity of delivery of a physical activityintervention: predictors and consequences. Psychol Health 2008;23:11–24.
47. Coulter A, Ellins J. Effectiveness of strategies for informing, educating and involvingpatients. BMJ 2007;335:24–7.
48. Bandura A. Social foundations of thought and action: a social cognitive theory.Englewood Cliffs, NJ, USA: Prentice Hall, 1986.
49. Carver CS, Scheier MF. Themes and issues in the self-regulation of behavior. In:Wyer RS, ed. Perspectives on behavioral self-regulation. London: Lawrence ErlbaumAssociates, 1999:1–105.
50. Hibbard JH, Stockard J, Mahoney ER, et al. Development of the Patient ActivationMeasure (PAM): conceptualising and measuring activation in patients andconsumers. Health Service Res 2004;39:1005–26.
51. Heneman K, Block-Joy A, Zidenberg-Cherr S, et al. A ‘‘contract for change’’increases produce consumption in low income women: a pilot study. J Am DieteticAssoc 2005;105:1793–6.
52. Mayer JA, Jermanovich A, Wight BL, et al. Changes in the health behaviours ofolder adults: the San Diego Medicare Preventive Health Project. Prev Med1994;23:127–33.
53. Kelley K, Abraham C. RCT of a theory-based intervention promoting healthy eatingand physical activity amongst out-patients older than 65 years. Soc Sci Med2004;59:787–97.
54. Glasgow RE, Toobert DJ, Hampson SE, et al. Implementation, generalization andlong-term results of the ‘‘choosing well’’ of diabetes self-management intervention.Patient Educ Counselling 2002;48:115–22.
55. Dombrowski S, Sniehotta FF, Avenel A, et al. Identifying active ingredients incomplex behavioural interventions for obese adults with additional risk factors: asystematic review. Under review.
APPENDICES
A. Databases searchedThe following databases were searched from January 1995 to September 2006:c ASSIA (Applied Social Sciences Index and Abstracts)c BiblioMapc CCTR (Cochrane Controlled Trials Register)c CDP (Chronic Disease Prevention)c CDSR (Cochrane Database of Systematic Reviews)c CHID (Combined Health Information Database)c CINAHL (Cumulative Index for Nursing and Allied Health Literature)c DARE (Database of Abstracts of Reviews of Effectiveness)c Dissertation Abstractsc Econlit (Economic Literature)c EMBASEc GreyLit (British Library grey literature collection)c HealthStarc HMIC (Healthcare Management Information Consortium)c MEDLINEc NHS EED (NHS Economic Evaluation Database)c Popline (Population Health and Family Planning)c PsychInfoc Social Science Citation Indexc Social Policy and Practicec Sociofile
B. Example search strategiesASSIA 220906Query(CSA) ((KW = (socioeconomic or socio-economic) and KW = (status or factor*))OR (KW = ((low income) or poverty or disadvantage*) or KW = deprive*)) AND(KW = (lifestyle* or activit* or inactivit*) or KW = ((weight gain) or overweight orobes*) or KW = ((food habit*) or (poor diet*))) AND ((KW = ((healthy eating) or fruit*or vegetable*) or KW = (exercis* or (health behaviour*) or (Health behavior*))) OR(KW = (interven* or Intervention*)))
Cinahl (Ovid) 2209061. socioeconomic factor$.ti,ab.2. exp Socioeconomic Status/3. socioeconomic status.ti,ab.4. (disadvantage$ or depriv$ or poverty or low income).ti,ab.5. (SES and status and low).ti,ab.6. or/1–57. exp LIFESTYLE/
8. lifestyle$.ti,ab.9. (activit$ or inactivit$).ti,ab.10. ((weight and gain) or overweight).ti,ab.11. exp OBESITY/12. obesity.ti,ab.13. food habit$.ti,ab.14. poor diet.ti,ab.15. or/7–1416. healthy eating.ti,ab.17. (fruit or vegetable$ or exercise$).ti,ab.18. exp Health Behavior/19. health behavior.ti,ab.20. health behaviour.ti,ab.21. health behaviour.ti,ab.22. (intervene$ or intervention$).ti,ab.23. or/16–2224. 6 and 15 and 2325. limit 24 to (english language and yr = ‘‘1996–2007’’)
Embase 2209061. socioeconomic factor$.ti,ab.2. exp Socioeconomic Status/3. socioeconomic status.ti,ab.4. (disadvantage$ or depriv$ or poverty or low income).ti,ab.5. (SES and status and low).ti,ab.6. or/1–57. exp LIFESTYLE/8. lifestyle$.ti,ab.9. (activit$ or inactivit$).ti,ab.10. ((weight and gain) or overweight).ti,ab.11. exp OBESITY/12. obesity.ti,ab.13. food habit$.ti,ab.14. poor diet.ti,ab.15. or/7–1416. healthy eating.ti,ab.17. (fruit or vegetable$ or exercise$).ti,ab.18. exp Health Behavior/19. health behavior.ti,ab.20. health behaviour.ti,ab.21. health behaviour.ti,ab.22. (intervene$ or intervention$).ti,ab.23. or/16–2224. 6 and 15 and 2325. limit 24 to (english language and yr = ‘‘1996–2007’’)
Medline (Ovid) 2209061. socioeconomic factor$.ti,ab.2. exp Socioeconomic Status/3. socioeconomic status.ti,ab.4. (disadvantage$ or depriv$ or poverty or low income).ti,ab.5. (SES and status and low).ti,ab.6. or/1–57. exp LIFESTYLE/8. lifestyle$.ti,ab.9. (activit$ or inactivit$).ti,ab.10. ((weight and gain) or overweight).ti,ab.11. exp OBESITY/12. obesity.ti,ab.13. food habit$.ti,ab.14. poor diet.ti,ab.15. or/7–1416. healthy eating.ti,ab.17. (fruit or vegetable$ or exercise$).ti,ab.18. exp Health Behavior/19. health behavior.ti,ab.20. health behaviour.ti,ab.21. health behaviour.ti,ab.22. (intervene$ or intervention$).ti,ab.23. or/16–2224. 6 and 15 and 2325. limit 24 to (english language and yr = ‘‘1996–2007’’)
PsycInfo (Ovid) 2209061. socioeconomic factor$.ti,ab.2. exp Socioeconomic Status/3. socioeconomic status.ti,ab.
Essay
J Epidemiol Community Health 2009;63:610–622. doi:10.1136/jech.2008.078725 621
on 16 July 2009 jech.bmj.comDownloaded from
4. (disadvantage$ or depriv$ or poverty or low income).ti,ab.5. (SES and status and low).ti,ab.6. or/1–57. exp LIFESTYLE/8. lifestyle$.ti,ab.9. (activit$ or inactivit$).ti,ab.10. ((weight and gain) or overweight).ti,ab.11. exp OBESITY/12. obesity.ti,ab.13. food habit$.ti,ab.14. poor diet.ti,ab.15. or/7–1416. healthy eating.ti,ab.17. (fruit or vegetable$ or exercise$).ti,ab.18. exp Health Behavior/19. health behavior.ti,ab.20. health behaviour.ti,ab.21. health behaviour.ti,ab.22. (intervene$ or intervention$).ti,ab.23. or/16–2224. 6 and 15 and 2325. limit 24 to (english language and yr = ‘‘1995–2007’’)
C. Previously unobtained full-text papers (n = 21)Full copies now obtained—all independently screened andexcluded (n = 6)1. Croghan IT, O’Hara MR, Schroeder DR, et al. A community-wide smoking
cessation program: quit and win 1998 in Olmsted County. Prev Med2001;33:229–38.
2. Fernandez E, Schiaffino A, Borrell C, et al. Social class, education, and smokingcessation: long-term follow-up of patients treated at a smoking cessation unit.Nicotine Tob Res 2006;8:29–36.
3. Freels SA, Warnecke RB, Parsons JA, et al. Characteristics associated withexposure to and participation in a televised smoking cessation intervention programfor women with high school or less education. Prev Med 1999;28:579–88.
4. Lazev AB, Vidrine DJ, Arduino RC, et al. Increasing access to smoking cessationtreatment in a low-income, HIV-positive population: the feasibility of usingcellular telephones. Nicotine Tob Res 2004;6:281–6.
5. Swinburn BA, Caterson I, Seidell JC, et al. Diet, nutrition and the prevention ofexcess weight gain and obesity. Public Health Nutr 2004;7:123–46.
6. Widga AC, Lewis NM, Fada RD. Defined, in-home, pre-natal nutritionintervention for low-income women. J Am Diet Assoc 1999;99:1058–62.
Unable to obtain full-text copies (n = 9)1. Cason KL, Scholl JF, et al. A comparison of program delivery methods for low
income nutrition audiences. Topics Clin Nutr 2002;17:63–73.
2. Finkelstein EA, Troped PJ, et al. Cost-effectiveness of a cardiovascular diseaserisk reduction program aimed at financially vulnerable women: theMassachusetts WISEWOMAN Project. J Women’s Health Gender-Based Med2002;11:519–26.
3. Foley RM, and Pollard CM. Food Cent$—implementing and evaluating anutrition education project focusing on value for money. Aust NZ J Public Health1998;22:494–501.
4. Freudenberg N, Silver D, et al. Health promotion in the city: a structured reviewof the literature on interventions to prevent heart disease, substance abuse,violence and HIV infection in US metropolitan areas, 1980–1995. J UrbanHealth 2000;77:443–57.
5. Kisioglu AN, Aslan B, et al. Improving control of high blood pressure amongmiddle-aged Turkish women of low socio-economic status through public healthtraining. Croatian Med J 2004;45:477–82.
6. Palmer TA, and Jaworski CA. Exercise prescription for underprivilegedminorities. Curr Sports Med Rep 2004;3:344–8.
7. Shi L. Sociodemographic characteristics and individual health behaviors.Southern Med J 1998;91:933–41.
8. Walcott-McQuigg JA. Psychological factors influencing cardiovascular riskreduction behavior in low and middle income African-American women. J NatlBlack Nurses’ Assoc 2000;11:27–35.
9. Will JC, Massoudi B, et al. Reducing risk for cardiovascular disease in uninsuredwomen: combined results from two WISEWOMAN projects. J Am MedWomens Assoc 2001;56:161–5.
Dissertations—excluded (n = 6)1. Davies SL. Using the transtheoretical model of change as a framework for
understanding smoking behavior in a sample of low-income, hospitalized,African-American smokers. Dissertation Abstracts International: Section B: TheSciences and Engineering 1997;58(5-B).
2. De Vogli R. Socioeconomic determinants of healthy lifestyles: does psychosocialstress matter? Dissertation Abstracts International: Section B: The Sciencesand Engineering 2004;65(2-B).
3. Dutton GR. Effects of a primary care weight management intervention onphysical activity in low-income African American women. DissertationAbstracts International: Section B: The Sciences and Engineering 2006;66(7-B).
4. Rounds TJ. Evaluation of a community-based home visiting program for lowincome families. Dissertation Abstracts International Section A: Humanities andSocial Sciences 1997;57(8-A).
5. Roundtree WJ. An analysis of parental scaffolding among three African-American mother–child dyads participating in the home instruction program forpreschool youngsters (HIPPY). Dissertation Abstracts International Section A:Humanities and Social Sciences 2000;61(6-A).
6. Springer J. Health behavior change as it relates to the adoption of andadherence to a program of physical activity. Dissertation AbstractsInternational: Section B: The Sciences and Engineering 2005;65(7-B).
Essay
622 J Epidemiol Community Health 2009;63:610–622. doi:10.1136/jech.2008.078725
on 16 July 2009 jech.bmj.comDownloaded from