obesity reviews doi: 10.1111/j.1467-789X.2012.01058.

Obesity Prevention/Management

Effectiveness of interventions to promote physical

activity among socioeconomically disadvantaged
women: a systematic review and meta-analysis

V. Cleland1,2, A. Granados1, D. Crawford1, T. Winzenberg2 and K. Ball1

1
Centre for Physical Activity and Nutrition
Research, Deakin University, Burwood,
Victoria, Australia; 2Menzies Research Institute
Tasmania, University of Tasmania, Hobart,
Tasmania, Australia
Received 11 July 2012; revised 13 September
2012; accepted 28 September 2012
Address for correspondence: Dr V Cleland,
Menzies Research Institute Tasmania,
University of Tasmania, Private Bag 23,
Hobart, Tasmania 7000, Australia
E-mail: verity.cleland@utas.edu.au
Summary
Physical activity is important for preventing weight gain and obesity, but women
experiencing socioeconomic disadvantage are at high risk of inactivity. This study
aimed to determine the effectiveness of interventions to increase physical activity
among women experiencing disadvantage, and the intervention factors (i.e. physi-
cal activity measure, delivery mode, delivery channel, setting, duration, use of
theory, behavioural techniques, participant age, risk of bias) associated with
effectiveness. We conducted a meta-analysis of controlled trials using random-
effects models and meta-regression. Seven databases were searched for trials
among healthy women (1864 years), which included a physical activity inter-
vention, any control group, and statistical analyses of a physical activity outcome
at baseline and post-intervention. Nineteen studies were included (n = 6,339).
Because of substantial statistical heterogeneity (c2 = 53.61, df = 18, P < 0.0001,
I2 = 66%), an overall pooled effect is not reported. In subgroup analyses, between-
group differences were evident for delivery mode, which modestly reduced het-
erogeneity (to 54%). Studies with a group delivery component had a standardized
mean difference of 0.38 greater than either individual or community-based deliv-
ery. Programs with a group delivery mode significantly increase physical activity
among women experiencing disadvantage, and group delivery should be consid-
ered an essential element of physical activity promotion programs targeting this
population group.

Keywords: Behaviour, female, prevention, socioeconomic factors.

obesity reviews (2012)

treating obesity-related disorders will increase by $US

Introduction
Globally, physical inactivity is the fourth leading risk factor
for mortality, and accounts for 6.6% and 7.7% of deaths in
middle- and high-income countries, respectively (1). Physi-
cal inactivity also substantially and independently increases
the risk of a vast range of common and chronic conditions,
such as cardiovascular diseases and certain cancers (2,3). If
current upward trends in obesity continue in the United
States (US), it is estimated that the annual medical costs for
2228 billion per year by the year 2020 and $US 4866
billion per year by the year 2030; respective figures for the
United Kingdom (UK) are 613648 million per year and
1.92 billion (4). The costs of physical inactivity have
been estimated at 0.9 billion in the UK (5), $AU 1.5
billion per year in Australia (6), around $US 3 billion in the
US (7) and $CAN 5.3 billion in Canada (8).
Despite the substantial social, economic and health
burdens of inactivity (2,3), a considerable proportion of

2012 The Authors 1

obesity reviews 2012 International Association for the Study of Obesity
2 Activity interventions on low SES women V. Cleland et al. obesity reviews

adults in industrialized countries are not physically active ity in women experiencing socioeconomic disadvantage,
at levels sufficient for good health (commonly accepted as and determine which, if any, intervention factors (i.e. physi-
150 min/week of at least moderate-intensity activity on cal activity measure, delivery mode, delivery channel,
most days) (9). Furthermore, the prevalence of physical setting, duration, use of theory, theory employed, number
inactivity is disproportionately distributed across popula- of behavioural techniques employed, mean participant age
tion segments. Women are less active than men across the and risk of bias) are associated with greater effectiveness.
lifespan (2,10), and there are clear socioeconomic inequali-
ties in leisure time physical activity participation, with
those with lower levels of education, in lower status Methods
occupations, and living in socioeconomically disadvan-
We performed a systematic review and meta-analysis of
taged areas demonstrating the lowest rates of physical
studies of interventions to increase physical activity in
activity (2,10,11). Women experiencing socioeconomic
women experiencing socioeconomic disadvantage. We
disadvantage are therefore an important group to target in
adhered to the Preferred Reporting Items for Systematic
developing and evaluating interventions to promote physi-
Reviews and Meta-Analyses statement for reporting sys-
cal activity. There may be additional public health benefits
tematic reviews and meta-analyses of studies that evaluate
in targeting this population group, because womens physi-
healthcare interventions (Supporting Information Appen-
cal activity behaviours may influence the behaviours of
dix S1) (32,33).
other proximal population groups, such as their children
and families. For instance, mothers act in important gate-
keeper roles, and their participation in physical activity has
Inclusion/Exclusion criteria
been shown to have a direct influence on their childrens
physical activity behaviours, more so than that of fathers Studies were eligible for inclusion if they met the following
(1215). criteria:
Despite this, the most effective strategies for promoting
Population group: community-dwelling women, or
physical activity among socioeconomically disadvantaged
studies where more than 80% of participants were women;
women are unknown. While numerous reviews have sum-
aged 1964 years, or studies with participants with a mean
marized the effectiveness of physical activity interventions
age <65 years; no pre-existing medical condition; targeted
(1630), none have directed their attention towards socio-
towards a population group experiencing socioeconomic
economically disadvantaged women who, due to their cir-
disadvantage (i.e. those with low education, low income,
cumstances, may have unique needs and barriers. Most
unemployed, low status occupations or living in an area of
reviews have not examined men and women separately
low socioeconomic status).
despite the known differential influences on physical activ-
Intervention: any intervention (individually, socially,
ity according to sex (31), have not included a meta-analysis
environmentally or policy targeted) focused on increasing
(1625,2729), and have not examined population groups
physical activity in any setting.
experiencing socioeconomic disadvantage, despite these
Control group: any control group including but not
groups being identified in a seminal review (22) as a key
limited to no intervention/contact, attention control or wait
target group about who little is known regarding effective
list control.
physical activity promotion strategies. Most previous
Outcome: outcomes measured, at a minimum, at base-
reviews have been unable to make firm conclusions about
line and immediately post intervention; reporting of statis-
the effectiveness of interventions on physical activity,
tical analyses of a physical activity outcome measure, or an
largely due to inconsistent findings, insufficient numbers of
outcome considered to be closely related to physical activ-
studies, concerns about study methodological quality, and
ity such as cardiorespiratory fitness.
the lack of use of meta-analytic techniques to quantify
Study design: randomized controlled trials and non-
effectiveness and the factors impacting on effectiveness.
randomized trials with a control group.
Reviews that have reported some favourable intervention
effects have tended to focus on specific population groups Studies were excluded from the review if they targeted
(e.g. children and adolescents) (2729), specific settings pregnant women, athletes or sports students; did not focus
(e.g. primary care) (22,26) or specific delivery channels (e.g. primarily on increasing physical activity; did not report on
mass media) (19). a physical activity outcome; if the sample included >20%
There are currently no data demonstrating which factors men and did not report a physical activity outcome sepa-
might improve the likelihood of an intervention being effec- rately for men and women; or only reported sedentary
tive among socioeconomically disadvantaged women. The behaviour as an outcome. Only studies with an outcome
aim of this systematic review was therefore to determine measure with a measure of variance available were
the effectiveness of interventions to increase physical activ- included in the review and meta-analysis.

2012 The Authors

obesity reviews 2012 International Association for the Study of Obesity
obesity reviews
Search strategy
The electronic search was last updated in March 2011. The
detailed search strategy is detailed in Supporting Informa-
tion Appendix S2. We searched: Pubmed, EMBASE,
Medline with Full Text, CINAHL, PsycINFO, Web of
Science and Global Health. The following search terms
were used: physical activity, exercise, health behaviour,
walking, socioeconomic disadvantage, disadvantaged,
underserved populations, women, female, intervention
studies, programs, health promotion programs. In addi-
tion, a manual citation search of the reference list of the
included studies and of two key physical activity journals,
the Journal of Physical Activity and Health, and the
International Journal of Behavioral Nutrition and
Physical Activity, was conducted. For included studies,
study authors were contacted when more information was
required; unpublished data were obtained from five authors
using this method.
Two reviewers (V.C. and A.G.) independently assessed
potentially relevant articles against the inclusion and exclu-
sion criteria. When results differed between reviewers, con-
sensus was reached through discussion; where consensus
could not be reached, a third author (K.B.) adjudicated.
Data extraction and analysis
A purpose-designed Microsoft Excel spreadsheet was
developed for data extraction (Supporting Information
Appendix S3). The spreadsheet was pilot tested using an
article retrieved but excluded from this review during the
search process, and refined accordingly. Data extraction
was performed by one reviewer (A.G.) and verified by a
second reviewer (V.C.). In cases of disagreement, discus-
sion was held until consensus was reached. The primary
outcome measure was any continuous measure of physical
activity, measured at the first time point at or after the
conclusion of the study. If these data were not available,
the proportions of participants meeting a specified level of
physical activity (defined by study authors, typically per-
forming 150 min/week of at least moderate-intensity activ-
ity) were extracted. Where multiple measures of physical
activity were included, we used an a priori hierarchy to
determine the appropriate measure for data extraction
(Supporting Information Appendix S3). In one study
where there was one control and two quite similar inter-
vention groups, the intervention group most disparate
from the control group was selected for inclusion in the
meta-analysis.
Risk of bias assessment
Assessing risk of bias in systematic reviews and meta-
analyses is essential because it impacts on the extent to
2012 The Authors
obesity reviews 2012 International Association for the Study of Obesity
Activity interventions on low SES women V. Cleland et al. 3
which conclusions can be drawn about intervention effects
(34). Risk of bias can also help to explain variation (het-
erogeneity) in the results of a meta-analysis, with more
rigorous studies being more likely to produce results that
are closer to the truth than less rigorous studies (34). Risk
of bias assessment of included articles was conducted using
the six-item Effective Public Health Practice Projects
Quality Assessment Tool for Quantitative Studies (35) (see
http://www.ephpp.ca/images/PDF/QATool.pdf). This tool
is recommended by the Cochrane Collaboration Australia
(34) (see http://www.cochrane-handbook.org for reviews
of public health and health promotion studies). The tool
considers selection bias, study design, confounders,
blinding, data collection methods, and withdrawals and
dropouts. Two reviewers (V.C. and A.G.) independently
assessed each of six risk of bias indicators as low,
medium or high; where consensus could not be reached,
a third author (K.B.) adjudicated. Studies were then given a
global risk of bias rating of low (four high ratings and no
low ratings), medium (less than four high ratings and one
low rating) or high (two or more low ratings from the six
components). Discrepancies were discussed until consensus
was reached. The level of agreement between the two
reviewers ratings was assessed using per cent agreement
and Cohens kappa coefficient (k).
Meta-analysis
Because measurement scales differed across studies, con-
tinuous physical activity outcomes were converted to
standardized mean differences (SMDs) and standard errors
(SEs). If continuous outcomes were not available, we used
dichotomous outcomes, re-expressing odds ratios and their
95% confidence intervals (CIs) as SMD and SE as described
below. For seven studies that presented the proportion of
participants in each group meeting a set level of physical
activity (e.g. physical activity recommendations), odds
ratios and 95% CIs were calculated then also re-expressed.
Based on an assumption that the underlying continuous
measurements in each intervention group follow a logistic
distribution, and that the variability of the outcomes is the
same in both treated and control participants, odds ratios
were re-expressed as SMDs using the following formula
(36):
3
SMD = ln OR
The SE of the log odds ratio was converted to the SE of a
SMD by multiplying with the same constant (3/
p = 0.5513).
Statistical heterogeneity was calculated using a chi-
squared test on N-1 degrees of freedom (df), with signifi-
cance conservatively set at 0.10. Inconsistency was also
4 Activity interventions on low SES women V. Cleland et al.
assessed using the formula, I2 = [(Qdf)/Q] 100%, where
Q is the chi-squared statistic and df is its degrees of
freedom, to describe the percentage of the variability in
effect estimates due to heterogeneity. A value greater than
50% was considered to be substantial heterogeneity
(37).
As there was substantial clinical and methodological
variation in the studies, meta-analysis was conducted using
a random-effects model. Our intent was that if statistical
heterogeneity were substantial, we would proceed to
exploring heterogeneity and identifying potential effect
modifying factors through pre-specified subgroup analyses
and random-effects meta-regression. These subgroups were
by: measure of physical activity (objective, valid/reliable
self-report, not valid/reliable self-report), delivery mode
(whether delivered only to individuals, only to groups, a
combination of group and individual delivery, or whole-of-
community approaches), delivery channel (face-to-face, tel-
ephone, mass media, print), setting (in the home, through
an organization/centre, or at the broader community
level), intervention duration (<6 months, 612 months, >12
months), use of a theoretical framework (yes, no), type of
theoretical framework, number of behavioural techniques
employed (based on the theory-linked Abraham taxonomy,
which provides standardized definitions of 26 behaviour
change techniques; two authors independently coded the
behaviour change techniques employed in each of the
included studies, with any discrepancies discussed until
consensus was reached) (38), mean age of participants and
risk of bias. Where possible, analyses were based on
intention-to-treat data from the individual trials. The
number of studies restricted meta-regression to a univari-
able approach. The SMD was used to estimate an absolute
benefit in units of min/week moderate- to vigorous-
intensity physical activity and steps/day by estimating the
pooled standard deviation (SD) from the means of the SD
of this outcome in treatment and control groups for each
study using this outcome, and multiplying the SMD by
this (39).
A funnel plot was generated for assessment of publica-
tion bias. Sensitivity analyses were performed to assess the
impact of risk of bias, of the use of non-validated self-
report outcome measures, of not appropriately adjusting
for clustering, and of the inclusion of studies with dichoto-
mous outcomes. To do this, analyses were rerun excluding
studies with a high risk of bias, studies that used a non-
validated self-report outcome measure, studies that did
not adjust for clustering, and studies that included only
dichotomous outcomes, respectively.
All analyses were performed in Review Manager
(Revman V5.0, Copenhagen: The Nordic Cochrane Centre,
The Cochrane Collaboration, 2008) except for meta-
regression that was performed in Stata Version 12 (College
Station, TX, USA).
obesity reviews
Results
Article search
Details of the results of the search strategy are presented in
Fig. 1. After the removal of 244 duplicates, the initial
search resulted in the identification of 1,695 articles. Arti-
cles were excluded (n = 1,257) where it could be ascer-
tained from the title that they did not meet the inclusion
criteria (described below). The abstracts of the remaining
articles (n = 438) were reviewed, and were excluded
(n = 343) if they did not meet the inclusion criteria. The full
texts of the remaining 95 articles were then retrieved and
assessed, and a further 78 articles were excluded because
they did not fit the inclusion criteria. A manual citation
search of the reference list of the remaining 17 studies and
of two physical activity journals was performed, with a
further two articles identified for inclusion, resulting in a
total of 19 studies eligible for inclusion in the review.
Summary of included studies
Nineteen separate studies were included in the systematic
review and meta-analysis. Characteristics of studies
included in the meta-analysis are given in Table 1. Of the
studies included in the meta-analysis, most were conducted
in North America (n = 10) (4049) or Europe (n = 4) (50
53), with a smaller number in Australia (n = 3) (5456),
Iran (n = 1) (57) and Colombia (n = 1) (58). Studies
were conducted between 1990 and 2010, and ranged in
sample size from 43 (48) to 1,578 (53) participants
(median = 108). Intervention duration ranged from 6
weeks (48) to 6 years (51) (median = 5 months). Thirteen
studies employed at least one theoretical framework, with
the most common being the transtheoretical model of
behaviour change (n = 4 studies) (40,42,44,57) and social
cognitive theory (n = 6 studies) (40,4446,55,58). Physical
activity was most commonly measured via self-report
(n = 16 studies). Twenty-one of the 26 behaviour change
techniques identified by Abraham and colleagues (38) were
employed (Supporting Information Appendix S4), with the
most commonly reported being providing information
about behaviourhealth links (n = 12 studies), prompting
of barrier identification (n = 11 studies), and planning for
social support or social change (n = 10 studies). None of
the studies reported adverse events or included a cost-
effectiveness component.
Risk of bias assessment
The per cent agreement between the two study raters was
83% for the global risk of bias assessment score (k = 0.65,
P < 0.001); full consensus was reached after discussion and
deferral to a third author was not required. Overall, 2
2012 The Authors
obesity reviews 2012 International Association for the Study of Obesity
obesity reviews
Articles identified
n = 1939
Articles initially reviewed by title
n = 1695
Articles eligible for review by abstract
n = 438
Articles eligible for review by full article
n = 95
Articles initially eligible for full review
n = 17
Hand-search identified articles
Articles eligible for full review
n = 19
Figure 1 Flow chart summary of articles identified in search and included in the review. aP: population group; I: intervention; C: control group;
O: outcome; S: study design; RCT: randomized controlled trial.
(11%) studies in the meta-analysis had low risk of bias, 3
(16%) had medium risk of bias, and 14 (74%) had high
risk of bias (Table 2). Because of the small number of
studies with low and medium risks of bias, these two cat-
egories were combined for meta-analyses.
Seven (37%) studies (40,43,44,52,55,56) applied or con-
sidered intention-to-treat analyses, and 12 (63%) studies
(4043,4648,52,54,56,57) reported the number of and/or
reasons for dropouts and withdrawals. Eight (42%) studies
identified no important differences between control and
intervention groups at baseline (40,4244,49,54,55,57),
two (11%) studies did not present this information (48,58),
and of the nine (47%) studies that found differences, four
2012 The Authors
obesity reviews 2012 International Association for the Study of Obesity
Activity interventions on low SES women V. Cleland et al. 5
Duplicates excluded
n = 244
Excluded by title
n = 1257
a
Excluded by abstract (n = 343)
P: not disadvantaged women (66)
I: not focused on physical activity (7)
C: no control group (78)
O: not physical activity (49)
S: not RCT or non-randomized with control (143)
Excluded by full article (n = 78)a
P: not disadvantaged women (61)
I: not focused on physical activity (3)
C: no control group (2)
O: not physical activity (7)
S: not RCT or non-randomized with control (5)
n=2
(21%) controlled for or considered controlling for these in
analyses (46,47,52,53). Only two (11%) studies (41,45)
clearly reported whether the consistency of the delivery of
the intervention was measured.
Eight (42%) studies (4143,47,48,55) described the
methods used for randomization sequence generation but
only one (43) described the methods used for allocation
concealment when this was applicable (i.e. randomized
controlled trial). In all studies, participants were not
blinded to the intervention status or research question (or
this information was unable to be discerned) (40,52), and
in only two (11%) studies were the personnel or outcome
assessors blinded to intervention status (43,45).
 6

Table 1 Summary of included studies examining the effectiveness of interventions to promote physical activity among socioeconomically disadvantaged women (n = 19)

Author (year), Target group Study design Setting (mode) Theoretical PA measure and Intervention details ITT Risk of
country and sample and duration framework summary results bias

Albright et al. 72 low-income, ethnic Randomized trial Home-based Social cognitive Self-report (V/R) energy All: Class-based component: 8 1 h weekly skill building classes; post-program Yes L
(2005) (40), minority sedentary (10-month trial; (telephone and theory and expenditure (kcal kg-1 d-1): assessments immediately after the generic home-based trial form the baseline data
USA women not exercising post-assessment mail) transtheoretical G0: Pre = 33.7 (SD: 2.2), G0: Home-based mail condition: standard health educational materials; monthly
as recommended: occurred 12 months model 12 m = 33.5 (SD: 1.5) newsletters; pedometers with no feedback.
G0: 31.8 years after baseline) G1: Pre = 33.2 (SD: 1.7), G1: Home-based phone + mail condition: systematic delivery of brief, structured PA
(SD: 10.8) 12 m = 33.2 (SD: 3.1) telephone counselling (1015 min) and stage of change-tailored newsletters
G1: 32.5 years (weekly for first 4 weeks, biweekly next 8 weeks, then monthly); pedometers with
(SD: 9.0) feedback and calendars; newsletters with mail back card to report steps, past
week minutes and types of PA (incentives provided for returned logs)
Baranowski 95 women from Randomized Centre based None stated Self-report G0: No contact control group No H
et al. (1990) Black-American controlled trial (face-to-face) (V/R) METs/week: G1: Education sessions: 90-min sessions: first 7 weeks, behavioural counselling,
(41), USA families: (14-week trial; G0: Pre = 235.5 (SD: 16.1), small group education, aerobic activity and healthy snack; last 7 weeks, sessions
G0: 32.9 years post-assessment 14 weeks = 248.0 (SD: 29.4) modified due to dropout (small group session dropped but information woven into
G1: 31.8 years occurred 14 weeks G1: Pre = 241.4 (SD: 22.8), individual family counselling, participants not required to attend education sessions
after baseline) 14 weeks = 247.8 (SD: 46.6) prior to fitness session and did not have to complete daily self-monitoring of diet
and exercise); participants encouraged to attend one education session and two
fitness sessions/week
Brown et al. 48 Greek-Australian Non-randomized Greek Orthodox None stated Objective (aerobic fitness) G0: No contact control group No H
(1996) (54), women: controlled trial church exercise HR (level 1 of G1: Weeks 112, participants attended 2-h weekly group meeting and followed
Australia G0: 47.0 years (24-week trial; (face-to-face) three-level test): home-based exercise program booklet; booklet material discussed (participants
(range: 2760) post-assessment G0: Pre = 103.5 (SD: 11.5), reported past week PA, identified barriers and solutions), followed by practical
G1: 46.8 years occurred 12 weeks 12 weeks = 98.7 (SD: 14.9) component (low impact, focus on assessing intensity and stretch safety); weeks
Activity interventions on low SES women V. Cleland et al.

(range: 3265) after baseline) G1: Pre = 114.2 (SD: 19.0), 1224, group meetings not held, participants worked independently on booklet
12 weeks = 98.5 (SD: 13.9)
Chang et al. 129 overweight and Randomized Community Social cognitive Self-report (V/R) METs/week: G0: ~20 min of nutrition education every 6 months No H
(2010) (45), obese controlled trial (media, theory and G0: Pre = 27.3 (SD: 29.9), G1: 5 1015 min DVD chapters biweekly plus five peer-supported group
USA AfricanAmerican and (10-week trial; telephone) formative 42 weeks = 36.0 (SD: 29.3) teleconferences (30 min/teleconference) at alternate weeks for 10 weeks.
Caucasian low-income post-assessment research G1: Pre = 29.8 (SD: 26.7), Participants answered quiz questions and set one to two goal(s) after each
mothers: occurred 8 months 42 weeks Post = 53.2 chapter; used weekly self-monitoring for 1 week; quiz and worksheets mailed to
G0: 25.1 years after trial completion, (SD: 30.2) study office.
(SD: 4.10) i.e. 42 weeks after
G1: 25.5 years baseline)
(SD: 3.94)
Fahrenwald 44 healthy, sedentary Randomized WIC community Transtheoretical Self-report (V/R) METs/day: G0: Provider counselling for self-breast examination (using brochure) plus four No M
et al. (2004) women with children controlled trial sites (telephone model G0: Pre = 32.59 (SD: 0.38), biweekly phone contacts (two revisiting self-breast exam and two on family health)
(42), USA enrolled in WIC: (8-week trial; and face-to-face) 10 weeks (change) = -0.17 G1: Moms on the Move provider counselling (1020 min) supplemented with
G0: 26.6 years post-assessment (SD: 0.41) brochure and four biweekly provider-delivered telephone contacts. Counselling
(SD: 6.8) occurred 2 weeks G1: Pre = 32.52 (SD: 0.39), involved identifying and discussing pros and cons to PA, barriers and strategies for
G1I: 25.2 years after trial completion, 10 weeks (change): 0.46 overcoming these; identifying personal PA goals, supportive others, type of support
(SD: 5.0) i.e. 10 weeks after (SD: 0.45) needed, and opportunities to counter sedentary habits. Telephone contact scripts
baseline) followed up on counselling messages.
Fjeldsoe et al. 88 post-natal women: Randomized Community Social cognitive Self-report (V/R) MVPA All: Initial face-to-face consultation with trained behavioural counsellor, standard Yes M
(2010) (55), G0: 31 years (SD: 6.0) controlled trial (face-to-face theory min/week: print-based PA information pack provided, including PA goal setting; baseline data
Australia G1: 28 years (SD: 6.0) (12-week trial; and telephone) G0: Pre = 84.0 (SE: 26.0), were collected after this session
post-assessment 13 weeks = 159.8 (SE: 29.3) G0: No additional contact or resources during trial period; reminder telephone calls
occurred 1 week after G1: Pre = 164.3 (SE: 25.4), to confirm 6- to 13-week assessments.
trial completion, i.e. 13 weeks = 149.8 (SE: 25.0) G1: 42 (35 SMS/week) personally tailored SMS providing behavioural and
13 weeks after cognitive strategies for behaviour change; 11 weekly goal check SMS and
baseline) instructions to nominate social support person. Participants set a short-term (6
weeks) PA goal during initial consultation, reviewed and updated at next phone
consultation. SMS tailored to each participants name, social support persons
name, youngest childs name, potential exercise partners name, PA goal,
nominated PA rewards and preferred PA options and signed off with first name of
behavioural counsellor.

2012 The Authors

obesity reviews

obesity reviews 2012 International Association for the Study of Obesity

 Table 1 Continued

Author (year), Target group Study design Setting (mode) Theoretical PA measure and Intervention details ITT Risk of

2012 The Authors

country and sample and duration framework summary results bias

Hovell et al. 151 low-income, Randomized Community Operant learning Self-report (% meeting PA G0: 18 90-min sessions (12 weekly, 6 bimonthly) over 6 months involved culturally Yes* M
(2008) (43), largely monolingual controlled trial (face-to-face) theory and guidelines): tailored, low-literacy education on home safety and selected disease prevention
obesity reviews

USA Spanish-speaking (6-month trial; applied G0: Pre = 13.6%, topics unrelated to exercise, diet or CVD
immigrant Latina post-assessment behaviour 12 m = 15.2% G1: Aerobic exercise component: 3 90-min group sessions/week for 6 months;
women: occurred 12 months analysis G1: Pre = 19.1%, involved vigorous low-impact aerobic dance, intensity/form monitoring and
31.4 years (SD: 6.2) after baseline) 12 m = 38.2% individualized feedback (10-min aerobic activity at week 1 was increased by
3 min/week up to 40 min; thereafter, exercise intensity was gradually increased).
Education component: following exercise, 30 min of exercise and diet education
provided with culturally appropriate hands-on learning activities presented using
visuals, stories and skits related to component behavioural skills; third session each
week dispelled misconceptions and problem-solved barriers to PA or study
participation; cultural attitudes about weight, food and exercise discussed and
myths debunked.
Jacobs et al. 511 low-income Randomized Community Social cognitive Self-report (not V/R) PA G0: Usual follow-up services at discretion of health departments (basic nutrition and Yes H
(2004) (44), women: controlled trial (telephone and theory, relapse score PA counselling pamphlets provided)
USA G0: 59 years (12-month trial; mail) prevention (042, not very to very G1: 6 bimonthly mailed computer-tailored health messages and two telephone
G1: 59 years post-assessment theory and active): calls (health department staff); materials computer tailored to each participants
occurred 12 months transtheoretical G0: Pre = 12.68 (SD: goals, stage of change, knowledge, social support system, high-risk situations for

obesity reviews 2012 International Association for the Study of Obesity

after baseline) model 5.96); 12 m = 12.98 relapse and perceived benefits and barriers to behaviour change and
(SD: 6.96) maintenance; messages designed for low-literate, low-income adult population and
G1: Pre = 12.84 featured testimonials, expert advice, feedback on health behaviours, behavioural
(SD: 6.51); 12 m = 12.86 contracts, self-monitoring forms, social support cards, health tips, quizzes;
(SD: 6.69) telephone calls assessed whether goals were being met and helped identify and
negotiate barriers and set new goals; separate materials were computer generated
for each participant based on individual pretest survey responses (32 messages
selected from 649 messages); for telephone calls, 10 messages were selected
from 141 potential messages.
Lucumi et al. 97 caregivers Non-randomized Community Social cognitive Self-report (V/R) walk for G0: Information and communication component involving 16 weekly 2-h sessions No H
(2006) (58), (women): controlled trial (face-to-face and theory transport or recreation provided as 8 sessions together with smoke-free home sessions; home caregivers
Colombia 37.3 years (5-month trial; telephone) 150 min/week (%): received materials about benefits of PA, how to maximize PA and how to use
(SD: 10.5) post-assessment G0: Pre = 5.3, 7 m = 5.3 places and opportunities of their daily routines to be active; communication
occurred 7 months G1: Pre = 27.8, 7 m = 33.3 component included group discussions about enablers and barriers to PA;
after baseline) educational component included stretching, dance classes and toning exercises,
and heart rate measurement.
G1: As per G0 plus brisk walking in the neighbourhood and a social support
component where key peers identified by women in these two groups were invited
to participate in the educational activities.
G2: As per G1 plus three telephone contacts to complete a questionnaire identifying
barriers to PA and assessing change of stage, and provision of guidelines for
overcoming barriers and moving to the next stage of change.
Lupton et al. 841 women fishermen Quasi-experimental Fishing Local Self-report (not V/R) % G0: No contact control No H
(2002) (51), and workers in the (6-year trial; community empowerment physically active: G1: The intervention started in 1988 (activities focused on improving working
Norway fishing industry: post-assessment (face-to-face) G0: Pre = 81.1%; conditions due to major fishing crisis); individual counselling was more strongly
G0: 47.9 years occurred 6 years 6 years = 83.2% emphasized after 1990 when advice was given about diet, smoking and physical
G1: 49.4 years after baseline) G1: Pre = 76.5%; activity as part of ordinary consultations with GPs, public health nurses and
6 years = 82.6% occupational health services; those with high myocardial infarction risk at 1987
screening (baseline assessment) received individual counselling
Activity interventions on low SES women V. Cleland et al. 7
 8

Table 1 Continued

Author (year), Target group Study design Setting (mode) Theoretical PA measure and Intervention details ITT Risk of
country and sample and duration framework summary results bias

Lupton et al. 656 F general Non-randomized Coastal Community Self-report (not V/R) % G0: Control group was matched to intervention group and did not receive the No H
(2003) (50), community members: controlled trial (3-year community empowerment physically active: intervention, but the countywide radio station covered the control communities and
Norway G0: 47.9 years trial; post-assessment (face-to-face and G0: Pre = 81.2%, the local newspaper distributed to one control community, and ~100 high-risk
G1: 47.5 years occurred 6 years mass media) 6 years = 80.9% individuals identified at baseline and intermediate screening were given individual
after baseline) G1: Pre = 73.0%, counselling
6 years = 80.9% G1: Intervention invited voluntary organizations to a workshop to identify barriers to
good health, and based on this, two groups developed a project manual with
health-promoting improvements; the manual was distributed to schools, a fish
factory, voluntary organizations and parts of the local public administration. ~270
providers from a range of voluntary organizations were involved in suggesting
changes and health-promoting activities regarding nutrition and PA. A second
project manual with all the suggestions listed in terms of short- and long-term
measures was discussed in meetings with the same groups, who were invited to
the workshop. Voluntary organizations carried out activities (e.g. the sports club
organized aerobic classes for ladies, physical training for individuals with heart
disease, badminton groups for grown-ups, cycle trails and was responsible for
preparing ski-tracks; the factory sports clubs association organized volleyball and
football tournaments; the pensioners association arranged dances; the rheumatic
association arranged swimming and special training in a heated pool). Guidelines
were established for individual counselling on exercise.
Olvera et al. 46 Latino Non-randomized Community and Social cognitive Self-report (PA rating; G0: Participants met with an instructor once a week for 90 min, received written No H
(2010) (46), motherdaughter controlled trial school theory 0 = sedentary, educational materials on various nutrition and counselling topics. Participants
Activity interventions on low SES women V. Cleland et al.

USA pairs (12-week trial; (face-to-face) 7 = vigorous): engaged in light-intensity aerobics (i.e. Samba) or sport sessions (i.e. basketball)
G0: 38.2 years post-assessment G0: Pre = 1.2 (SD: 1.5), for 45 min.
(SD: 10.6) occurred 12 weeks 12 weeks = 1.2 (SD: 0.9) G1: Three weekly structured group aerobics (e.g. Salsa) or sport sessions (e.g.
G1: 33.3 years after baseline) G1: Pre = 1.4 (SD: 0.9), basketball) or free-play recreational activities, two weekly nutrition sessions and 1
(SD: 4.6) 12 weeks = 2.1 (SD: 1.6) weekly behavioural counselling session. Each session included 45 min of exercise
and 45 min of nutrition education or counselling.
Opdenacker 169 rural women: Randomized Catholic rural None stated Objective (accelerometer) G0: No contact control Yes* H
et al. (2008) G0: 53.0 years controlled trial womens counts/5 d: G1: 1 intervention meeting, a self-help booklet, and monthly reminders about why
(52), Belgium (SD: 12.5) (6-month trial; organization G0: Pre = 1,664,013 (SD: and how to incorporate more PA into daily life; 90-min intervention meeting
G1: 54.5 years post-assessment (face-to-face and 521,275), 6 m = 1,501,413 consisted of (1) a presentation of benefits of regular PA and recommendations for
(SD: 12.0) occurred 6 months print) (SD: 594,714) moderate and vigorous PA; (2) goal setting exercises resulting in an individually
after baseline) G1: Pre = 1,702,474 (SD: tailored 6-month PA contract; (3) explanation of an exercise calendar for
618,907), 6 m = 1,827,888 self-monitoring; (4) interactive discussions on barriers, reminders and rewards; and
(SD: 687,279) (5) information about pedometers; self-help booklet consisted of 20 pages with
pictures and information similar to that provided during the meeting. Afterwards,
participants received five monthly reminder letters with encouragements and tips to
stay active, in which the essential information of the meeting and the self-help
booklet was repeated and elaborated.
Shirazi et al. 116 older Iranian Randomized Health centres Transtheoretical Self-report (V/R) PA G0: Wait list control No L
(2007) (57), women: controlled trial and homes model min/week: G1: Education program: a common component involved instructional strategies.
Iran G0: 52.8 years (12-week trial; (face-to-face) G0: Pre = 73.9 (SD: Participants were divided into seven subgroups according to stage of change and
(SD: 8.8) post-assessment 131.2), 12 weeks = 78.9 given tailored instructions over 2 subsequent weeks (lectures presented using
G1: 53.5 years occurred 12 weeks (SD: 136.2) videos, slides and posters, group discussion and self-estimation of osteoporosis
(SD: 7.9) after baseline) G1: Pre = 54.1 (SD: risk). Materials aimed to increase osteoporosis knowledge (personal susceptibility
131.5), 12 weeks = 191.4 and fracture risk); instructions were tailored to stage of change every 2 weeks.
(SD: 231.4) Exercise program: Began with individual home visit 30- to 45-min sessions for those
in preparation and action stages; participants shown exercise booklet; instructors
ensured participants understood exercises and were safe and confident. Exercise
was progressive and individually tailored, and separated by 1 d of rest. After
strengthening exercises, participants performed one of three balance progressive
training exercises; walking at least 30 min d-1 broken up into two to three 10-min
sessions was prescribed.

2012 The Authors

obesity reviews

obesity reviews 2012 International Association for the Study of Obesity

 Table 1 Continued

Author (year), Target group Study design Setting (mode) Theoretical PA measure and Intervention details ITT Risk of
country and sample and duration framework summary results bias

Speck et al. 104 low-income Non-randomized Church-sponsored Adapted Health Objective (pedometer) G0: No contact control No H
(2007) (49), women: controlled trial community Promotion Model steps/day: G1: ~6 PA opportunities offered each week, including 4 at community centre
USA 39.6 years (SD: (26-week trial; centre G0: Pre = 5,314.6 (exercise and weight rooms, low-level aerobic-type PA classes, and hip-hop), and

2012 The Authors

12.8) post-assessment (face-to-face, (SD: 2,862.5), 23 two neighbourhood walks (one with a community resident and one with the nurse);
occurred 23 weeks telephone and weeks = 4,094.9 nurse conducted telephone prompts to encourage participation, provided sessions
after baseline) mail) (SD: 2,735.9) on general health and PA, and led a community walk; two special activities were a
G1: Pre = 5,791.3 walking trip to a store to shop and read labels, and participation at a centre health
(SD: 2,995.4), 23 fair. Three newsletters were mailed to subjects over 6 months and included PA
obesity reviews

weeks = 5,369.6 information with an emphasis on incorporating PA into daily life.

(SD: 2,786.5)
Stoddard et al. 1,443 uninsured and Randomized WIC project None stated Self-report (% adequate G0: Breast and cervical cancer screening, CVD risk factor screening, onsite multiple No H
(2004) (47), underinsured women controlled trial sites PA): risk factor assessment, counselling and education, referrals and follow-up as
USA aged 50 years: (12-month trial; (face-to-face) G0: Pre = 45.8%, needed; included a health risk appraisal to assess blood pressure, cholesterol and
5064 years: 82.9%, post-assessment 12 m = 52.0% health behaviours, including diet, PA and smoking; low-literacy fact sheets on
65 years: 17.1% occurred 12 months G1: Pre = 36.4%, cholesterol, blood pressure, blood glucose, nutrition, PA and stress reduction.
after baseline) 12 m = 54.5% G1: As per G0 plus lifestyle interventions focused primarily on nutrition and PA to
reduce CVD risk, including one-on-one nutritional and PA assessments and
counselling, individual and group education, and behavioural intervention activities
(with goals to help learn new information and skills and gain support for integrating
healthy behaviours into their lives); each site offered a unique assortment of nutrition
and PA interventions that reflected the resources available in the community and the
creativity and initiative of participants and staff. Activities included walking groups,
nutrition classes on modifying or translating recipes, and cultural festivals.

obesity reviews 2012 International Association for the Study of Obesity

Watson et al. 108 post-partum Non-randomized Community None stated Self-report (% adequate G0: Wait list control Yes H
(2005) (56), women: controlled trial (face-to-face) PA**) G1: 1-h pram walking groups once a week in numerous locations across the
Australia G0: 29.2 years (6-month trial; G0: Pre = 22.9, 6 m = 35.4 intervention area; after 3 months all mothers were invited to attend a walking group
G1: 29.6 years post-assessment G1: Pre = 33.3, 6 m = 43.3 leaders training course with the aim of mothers taking over the leadership role
occurred 6 months during the last 2 months, in preparation for the project officers withdrawal
after baseline)
Wendel-Vos 1,578 women: Non-randomized Community None stated Self-report (V/R) LTPA G0: A control region 200 km from the intervention region received no intervention No H
et al. (2009) G0: 51.3 years (SD: controlled trial (5-year (print and h/week: G1: Encouraged people to reduce fat intake, be physically active and stop
(53), the 10.4) trial; post-assessment mass media) G0: Pre = 18.3 (SD: 12.8), smoking. The intervention was a large umbrella project in which 790 interventions
Netherlands G1: 50.6 years (SD: occurred 5 years 5 years = 17.4 (SD: 12.4) were implemented, of which 590 were major (193 dietary, 361 PA and 9
9.7) after baseline) G1: Pre = 15.4 (SD: 11.7), antismoking). Examples included nutrition parties; debt assistance (people with
5 years = 17.2 (SD: 12.9) debts were taught to cook healthy meals on a small budget); printed guides
showing walking and cycling routes; a daily television-guided aerobics program,
including information about the health advantages of exercise; and antismoking
campaigns using billboards, posters and leaflets. About 50% of these interventions
were in deprived areas.
Williams et al. 43 post-menopausal Randomized Community Andragogy Self-report (% meeting brisk G0: Written information about benefits of exercise and health risks of inactivity; No H
(2005) (48), AfricanAmerican controlled trial (face-to-face) walking goal): instruction in use of pedometers; counselling session to identify walking goals,
USA women: (6-week trial; G0: 6 weeks = 31% potential barriers, motivational strategies and rewards. One goal was achieving
57.6 years (range: post-assessment G1: 6 weeks = 81% 30 min d-1 of moderate-intensity exercise most days of the week; the other was a
5068), occurred 6 weeks pedometer goal, negotiated according to the baseline average daily steps, current
after baseline) recommendation of 10,000 steps/day and the participants suggested targeted value.
G1: As per G0 plus participants signed a contract specifying goals and were given a
copy of the contract.

*ITT conducted but results comparable so not reported.

Standard deviation not available.
Information on age only available for whole sample (not stratified by control/intervention group status).
In each of these studies, participants self-reported their physical activity that was then converted to energy expenditure or METs by the study authors using standard formula.
Adequate PA defined by the study authors as engaging in physical activity at least three times a week and reporting at least 30 min of moderate or vigorous activity on an average day.

**Adequate PA defined by the study authors as 150 min of activity accrued over the past week on at least five separate sessions of activity (with vigorous activity weighted by a factor of 2).
Prams (perambulators) are also known as strollers, baby carriages, pushchairs or buggies.

CVD, cardiovascular disease; G0, control/comparison group; G1, intervention group 1; GP, general practitioner; H, high risk of bias; HR, heart rate; ITT, intention-to-treat; L, low risk of bias; LTPA, leisure time physical activity; M, medium
Activity interventions on low SES women V. Cleland et al. 9

risk of bias; METs, metabolic equivalents; MVPA, moderate- to vigorous-intensity physical activity; PA, physical activity; SD, standard deviation; SE, standard error; SMS, short message service; V/R, valid/reliable; WIC, women, infants and
children.
10 Activity interventions on low SES women V. Cleland et al. obesity reviews

Table 2 Risk of bias assessment

Author (year) Selection Study Confounding Blinding Data collection Withdrawals Global risk
bias design methods and dropouts of bias rating*

Albright et al. (2005) (40) Moderate Low Low Moderate Low Low Low
Shirazi et al. (2007) (57) Moderate Low Low Moderate Low Low Low
Fahrenwald et al. (2004) (42) Moderate Low Low High Low Low Moderate
Fjeldsoe et al. (2010) (55) Moderate Low Low High Low Moderate Moderate
Hovell et al. (2008) (43) High Low Low Moderate Low Low Moderate
Baranowski et al. (1990) (41) Low Low High High High High High
Brown et al. (1996) (54) High Low Low High Low Low High
Chang et al. (2010) (45) Moderate Low High Moderate Low High High
Jacobs et al. (2004) (44) High Low Low High High Low High
Lucumi et al. (2006) (58) High Low High High Low Moderate High
Lupton et al. (2002) (51) Moderate Moderate High High High Moderate High
Lupton et al. (2003) (50) Moderate Moderate High High High Moderate High
Olvera et al. (2010) (46) High Low Low High Low Low High
Opdenacker et al. (2008) (52) High Low Low High Low High High
Speck et al. (2007) (49) High Low High Moderate Moderate Moderate High
Stoddard et al. (2004) (47) High Low Low High High Moderate High
Watson et al. (2005) (56) High High High High High Moderate High
Wendel-Vos et al. (2009) (53) High Moderate Low Moderate Low Low High
Williams et al. (2005) (48) High Low High High Low Low High

*Global risk of bias ratings is defined as: low (four low ratings and no high ratings), medium (less than four low ratings and one high rating), high
(two or more high ratings from the six components).

While most studies did not directly include reliability
information for the self-reported physical activity measures
in their reports, many made reference to other articles for
this information. Six (32%) studies (41,44,47,4951) pro-
vided no such reliability data or references of their self-
reported measures. Of those that did provide information
on reliability, this was generally tested using a 1- or 2-week
test-retest approach, with intra-class, Pearson or Spearman
correlations reported in the range of 0.300.95 (median
0.80). Seven (37%) studies (41,43,44,47,50,51,56) pro-
vided no information on validity of self-reported measures
of physical activity, while some studies examined predictive
validity (degree of association between physical activity
measure and a relevant health outcome) or convergent
validity (degree to which measure associated with other
similar measures of physical activity) (43,48,53). Of the
studies that compared self-reported physical activity to an
objective measure (such as accelerometer or pedometer)
(40,42,45,49,52,55,57,58), correlation coefficients ranged
from 0.17 to 0.87 (median 0.38).
Meta-analysis of effectiveness of interventions
The meta-analysis included data from 19 studies with a
total of 6,339 participants (4058). The initial random-
effects meta-analysis of all studies confirmed that there was
substantial statistical heterogeneity, as we expected from
the clinical and methodological heterogeneity observed in
the studies (c2 = 53.61, df = 18, P < 0.0001, I2 = 66%).
Thus, pooling of the overall result was inappropriate, and
thus the result of this analysis is not reported. We pro-
ceeded to explore heterogeneity and examine potential
effect modification through our pre-specified subgroup
analyses with meta-regression, the results of which are
given in Table 3. For delivery mode, subgroup analyses
demonstrated that studies using group and those using
group in combination with individual delivery modes had
similar effect sizes of SMD 0.40 (95% CI 0.140.67) and
0.32 (95% CI 0.050.59), respectively. These two groups
were reclassified into a category of any group component
for meta-regression as presented in Table 3.
The only variable that had significant between-group
differences was delivery mode, with studies that had a
group component having a SMD of 0.38 greater than indi-
vidual delivery alone, and 0.38 greater than community-
based delivery modes (Fig. 2). Meta-regression with this
explanatory factor resulted in a modest reduction in het-
erogeneity (residual heterogeneity 54.2% compared to
66% in pooled analysis of all studies). Residual heteroge-
neity was also modest for intervention duration (53.8%),
but no between-group differences were observed for this
variable.
For all other variables, subgroup analyses did not
explain statistical heterogeneity with residual heterogeneity
being 65% or more in all cases, and there were no signifi-
cant between-group differences in effects.

2012 The Authors

obesity reviews 2012 International Association for the Study of Obesity
obesity reviews Activity interventions on low SES women V. Cleland et al. 11

Table 3 Summary of analysis of heterogeneity

No. of SMD (95% CI) P for between-group Residual

studies difference heterogeneity

Measure of physical activity 65.0%

Objective 3 0.40 (0.13, 0.66) Ref
Self-report (valid/reliable) 12 0.25 (0.02, 0.48) 0.576
Self-report (not valid/reliable) 4 0.01 (-0.08,0.10) 0.166
Delivery mode 54.2%
Any group component 11 0.36 (0.17, 0.54) Ref
Individual 5 -0.02 (-0.35,0.31) 0.029
Community 3 -0.02 (-0.10,0.05) 0.026
Delivery channel 65.3%
Face-to-face 9 0.35 (0.10, 0.60) Ref
Face-to-face + other 4 0.12 (-0.33,0.57) 0.348
Telephone + mail 3 0.14 (-0.26,0.55) 0.100
Mass media + other 3 -0.02 (-0.10,0.05) 0.381
Setting 66.8%
Community 12 0.14 (-0.02,0.30) Ref
Organization/Centre 6 0.26 (0.03, 0.49) 0.571
Home* 1 -0.13 (-0.59,0.33) 0.472
Duration 53.8%
12 weeks 6 0.37 (-0.06,0.80) Ref
>12 weeks and <12 months 8 0.27 (0.05, 0.49) 0.727
12 months 5 -0.00 (-0.07,0.06) 0.081
Use of theory 66.8%
Yes 13 0.23 (0.03, 0.43) Ref
No 6 0.07 (-0.05,0.20) 0.523
Theory employed 68.6%
None 6 0.07 (-0.05,0.20) Ref
SCT 4 0.33 (-0.27,0.94) 0.505
TTM 2 0.24 (-0.51,0.98) 0.630
Multiple theories 2 -0.04 (-0.22,0.14) 0.548
Other theories 5 0.32 (-0.01,0.66) 0.379
Number of behavioural strategies 69.4%
<5 9 0.12 (-0.06,0.31) Ref
58 6 0.15 (-0.12,0.43) 0.934
>8 4 0.28 (-0.09,0.64) 0.654
Mean age (years) 68.7%
<35 8 0.17 (-0.12,0.45) Ref
3550 5 0.05 (-0.12,0.23) 0.875
>50 6 0.20 (0.01, 0.39) 0.525
Risk of bias 67.4%
High 14 0.15 (0.03, 0.27) Ref
Low/Medium 5 0.13 (-0.31,0.57) 0.773

Bold denotes statistically significant between-group level differences.

*Only one study in category.
CI, confidence interval; SCT, social cognitive theory; SMD, standardized mean difference; TTM, transtheoretical model.

cal activity measure employed, appropriate adjustment for

Sensitivity analyses
A funnel plot was generated for assessment of publication
bias (Fig. 3) and visual assessment of this suggested a
degree of asymmetry due to the absence of any small pub-
lished negative studies. This may indicate publication
bias, but a sensitivity analysis omitting the two studies of
Lucumi (58) and Williams (48) contributing to this asym-
metry did not materially affect the results (data not shown).
Sensitivity analyses of the impact of risk of bias, the physi-
clustering, and the use of dichotomous outcomes also iden-
tified no meaningful differences in results (data not shown).
Discussion
This first systematic review and meta-analysis of physical
activity interventions among women experiencing socio-
economic disadvantage demonstrates clearly that the key
factor determining intervention effectiveness is mode of

2012 The Authors

obesity reviews 2012 International Association for the Study of Obesity
12 Activity interventions on low SES women V. Cleland et al.
Figure 2 Forest plot of subgroup analysis of the impact of delivery mode on intervention effectiveness.
delivery, specifically the use of group-based interventions.
Interventions including this component had both a statis-
tically and clinically significantly greater effect than studies
using other modes of delivery, equating to achieving an
additional 70 min/week of physical activity or approxi-
mately 1,000 steps/day. The potential public health and
clinical importance of an effect of this magnitude is sub-
stantial, with this volume of physical activity making a
substantial contribution to the minimum amount required
(15 min d-1) to significantly reduce all-cause mortality (59).
Previous reviews have struggled to make firm statements
about the effectiveness of physical activity promotion inter-
ventions, and specifically about intervention factors that
contribute to intervention effectiveness (16,17,19,21
24,30). Therefore, our results make an important contri-
bution to informing the design of public health
obesity reviews
interventions targeting this high-risk population group,
and provide fundamental guidance for policy-makers and
service deliverers.
The type of formats in studies that incorporated a group
component generally consisted of group education meet-
ings (45,52,57), practical sessions (46,49,56,58) or a com-
bination of both (41,43,47,54), facilitated by a trained
educator, health worker or practitioner. The importance of
group-based interventions likely lies in the social support
mechanisms provided by group settings. Social support can
be operationalized in a number of ways, including instru-
mental support (e.g. assistance with transportation), infor-
mational (e.g. sharing of educational resources), emotional
(e.g. asking how a physical activity schedule or program is
going) or appraisal (e.g. encouraging or reinforcing activ-
ity) (60). Our work has identified social support from
2012 The Authors
obesity reviews 2012 International Association for the Study of Obesity
obesity reviews
Figure 3 Funnel plot to assess risk of
publication bias. The absence of studies on
the lower left-hand side of the plot in contrast
with two studies (48,58) on the lower
right-hand side suggests possible publication
bias.
family and friends as key independent predictor of physical
activity among women experiencing socioeconomic disad-
vantage (6163). Social support has also been consistently
identified as an important correlate of physical activity
behaviour in a range of population groups (6466), and
particularly among women (6769). In this review, we were
unable to extract information on the type of support pro-
vided or whether establishing new or tapping into existing
groups impacted on the effectiveness of interventions.
However, these considerations warrant further research
attention, as certain forms of social support may be easier
to enact than others, and the utilization of existing group
structures may represent a more feasible and sustainable
approach than establishing new groups.
Other explanations for the effectiveness of group-based
interventions in this population group may include a shared
group experience (e.g. sharing barriers and identifying
common solutions), opportunities for friendship outside
of the home environment, accountability (which we have
found previously to be an important consideration for
intervention development among women experiencing
socioeconomic disadvantage) (70) or feelings of collective
efficacy (e.g. striving towards a similar goal). Further
studies should explore these concepts in their process and
qualitative evaluation of interventions to determine which
factors may be most important for this population group.
Also worthy of further consideration is whether group
size, format (e.g. education alone, practical alone or a
combination) and type of facilitator impact on intervention
effectiveness.
Because of the diverse combinations of behavioural
techniques and theoretical frameworks employed across
studies, we were unable to examine whether specific types
or differing combinations of behavioural techniques were
2012 The Authors
obesity reviews 2012 International Association for the Study of Obesity
Activity interventions on low SES women V. Cleland et al. 13
more effective for promoting physical activity than others.
The number of behavioural techniques employed did not
impact intervention effectiveness; however, it was surpris-
ing to note that only one study included information on
time management, despite lack of time being identified
commonly and consistently as a barrier to physical activity,
including among those experiencing socioeconomic disad-
vantage (7173). For instance, one study found that 73%
of women identified lack of time as a key barrier to par-
ticipation in physical activity, with the main reasons for
lack of time for physical activity being long work/study
hours and family commitments (72). Another study of
women found 58% reported lack of time due to work
commitments as a major barrier to physical activity (74). It
is plausible that barrier identification strategies, which were
one of the most common behavioural techniques employed
in the studies in this review, tapped into time barriers.
However, given that lack of time appears to be such a
critical barrier for womens participation in physical
activity, interventions that incorporate time management
strategies may be warranted.
Despite more than 70% of the studies including a theo-
retical framework, this factor bore no relationship to the
effectiveness of the interventions. The application of theo-
retical frameworks in the design of behavioural interven-
tions has been strongly advocated (7577). While this
finding may suggest the use of theory as unimportant, how
the theoretical frameworks were operationalized in the
design and implementation of the included studies, and to
what level studies adhered to the selected theory/theories, is
unclear. Future work should clearly elucidate how theoreti-
cal frameworks are incorporated into the design and imple-
mentation of physical activity interventions, and process
evaluations should consider fidelity to the study design and
14 Activity interventions on low SES women V. Cleland et al.
overarching theoretical framework in their assessments, so
that the use of theory can be examined more thoroughly as
a predictor of intervention success.
Of note, not one study included in this review reported
on intervention cost-effectiveness, and none reported on
adverse outcomes. Further studies must include this infor-
mation in order to provide a convincing (or otherwise) case
to policy-makers and service deliverers.
Potential limitations
This review has some limitations. Most studies used self-
reported measures of physical activity, and the risk of bias
was high only 5 of the 19 studies were deemed to have
medium or low risk of bias. Further, a number of studies
did not account for clustering in their study design, and in
some cases we had to calculate SMDs and SEs from
dichotomous data. However, none of these factors mean-
ingfully altered findings in sensitivity analyses, and neither
physical activity measure nor risk of bias rating explained
heterogeneity in the meta-analysis, suggesting the likely
impact was minimal. There was substantial clinical, meth-
odological and statistical heterogeneity in the studies exam-
ined, which meant that a pooled SMD for all studies could
not be meaningfully estimated. However, our exploration
of this heterogeneity revealed a clear indication of an
important study design factor, namely the use of groups,
which influenced the effectiveness of interventions. We only
included published studies and those published in English.
However, no studies were actually excluded due to lan-
guage of publication, and seven databases were systemati-
cally searched twice (October 2009; March 2011),
supplemented by hand searching of reference lists and key
physical activity journals making our search strategy oth-
erwise very comprehensive. Publication bias cannot be
ruled out, although a large number of small unpublished
studies with negative findings would be required to have an
impact on the overall results.
Conclusions
This meta-analysis fills a substantial gap in the literature
about the effectiveness of physical activity interventions
among a population group at high risk for inactivity and
chronic disease. The findings clearly demonstrate the
importance of group-based programs to promote physical
activity among women experiencing socioeconomic disad-
vantage. Programs of this nature are likely to have a clini-
cally meaningful impact on physical activity behaviour and
are highly warranted among this population group. Studies
investigating the impact of type of social support and the
size, nature and format of group interventions to promote
physical activity will provide further important insights for
policy and practice.
obesity reviews
Conflict of Interest Statement
All authors declare that they have no competing interests.
Acknowledgements
V.C. is supported by a National Health and Medical
Research Council Public Health Training (Postdoctoral)
Fellowship. A.G. is supported by a National Health and
Medical Research Council Strategic Award. T.W. is sup-
ported by a National Health and Medical Research Council
/Primary Health Care Research, Evaluation and Develop-
ment Career Development Fellowship. K.B. is supported by
a National Health and Medical Research Council Senior
Research Fellowship. D.C. is supported by a Victorian
Health Promotion Foundation Senior Research Fellowship.
References
1. World Health Organization. Global Health Risks: Mortality
and Burden of Disease Attributable to Selected Major Risks.
World Health Organization: Geneva, 2009.
2. Haskell WL, Lee IM, Pate RR et al. Physical activity and public
health: updated recommendation for adults from the American
College of Sports Medicine and the American Heart Association.
Med Sci Sports Exerc 2007; 39: 14231434.
3. US Department of Health and Human Services. Physical Activ-
ity and Health: A Report of the Surgeon General. United States
Department of Health and Human Services, Centers for Disease
Control and Prevention, National Center for Chronic Disease Pre-
vention and Health Promotion: Atlanta, GA, 1996, pp. 135141.
4. Wang YC, McPherson K, Marsh T, Gortmaker SL, Brown M.
Health and economic burden of the projected obesity trends in the
USA and the UK. Lancet 2011; 378: 815825.
5. Scarborough P, Bhatnagar P, Wickramasinghe KK, Allender S,
Foster C, Rayner M. The economic burden of ill health due to diet,
physical inactivity, smoking, alcohol and obesity in the UK: an
update to 200607 NHS costs. J Public Health (Oxf) 2011; 33:
527535.
6. Medibank Private. The cost of physical inactivity. Medibank
Private, 2007.
7. Chenoweth D, Leutzinger J. The economic cost of physical
inactivity and excess weight in American adults. J Phys Act Health
2006; 3: 148163.
8. Katzmarzyk PT, Janssen I. The economic costs associated with
physical inactivity and obesity in Canada: an update. Can J Appl
Physiol 2004; 29: 90115.
9. World Health Organization. The World Health Report 2002:
Reducing Risks, Promoting Healthy Life. World Health Organi-
zation: Geneva, 2002.
10. Pate RR, Pratt M, Blair SN et al. Physical activity and public
health. A recommendation from the Centers for Disease Control
and Prevention and the American College of Sports Medicine.
JAMA 1995; 273: 402407.
11. Kavanagh AM, Goller JL, King T, Jolley D, Crawford D,
Turrell G. Urban area disadvantage and physical activity: a mul-
tilevel study in Melbourne, Australia. J Epidemiol Community
Health 2005; 59: 934940.
12. Spurrier NJ, Magarey AA, Golley R, Curnow F, Sawyer MG.
Relationships between the home environment and physical activity
2012 The Authors
obesity reviews 2012 International Association for the Study of Obesity
obesity reviews
and dietary patterns of preschool children: a cross-sectional study.
Int J Behav Nutr Phys Act 2008; 5: 31.
13. Cleland V, Timperio A, Salmon J, Hume C, Telford A, Craw-
ford D. A longitudinal study of the family physical activity envi-
ronment and physical activity among youth. Am J Health Promot
2011; 25: 159167.
14. Salmon J, Ball K, Crawford D et al. Reducing sedentary
behaviour and increasing physical activity among 10-year-old chil-
dren: overview and process evaluation of the Switch-Play inter-
vention. Health Promot Int 2005; 20: 717.
15. Jacobi D, Caille A, Borys JM et al. Parent-offspring correla-
tions in pedometer-assessed physical activity. Plos ONE 2011; 6:
e29195.
16. Taylor WC, Baranowski T, Young DR. Physical activity inter-
ventions in low-income, ethnic minority, and populations with
disability. Am J Prev Med 1998; 15: 334343.
17. Yancey AK, Kumanyika SK, Ponce NA et al. Population-based
interventions engaging communities of color in healthy eating and
active living: a review. Prev Chronic Dis 2004; 1: A09.
18. Kahn EB, Ramsey LT, Brownson RC et al. The effectiveness of
interventions to increase physical activity. A systematic review. Am
J Prev Med 2002; 22: 73107.
19. Marcus BH, Owen N, Forsyth LH, Cavill NA, Fridinger F.
Physical activity interventions using mass media, print media,
and information technology. Am J Prev Med 1998; 15: 362
378.
20. Conn VS, Minor MA, Burks KJ, Rantz MJ, Pomeroy SH.
Integrative review of physical activity intervention research with
aging adults. J Am Geriatr Soc 2003; 51: 11591168.
21. Cyarto EV, Moorhead GE, Brown WJ. Updating the evidence
relating to physical activity intervention studies in older people. J
Sci Med Sport 2004; 7: 3038.
22. Eakin EG, Glasgow RE, Riley KM. Review of primary care-
based physical activity intervention studies: effectiveness and
implications for practice and future research. J Fam Pract 2000;
49: 158168.
23. King AC, Rejeski WJ, Buchner DM. Physical activity interven-
tions targeting older adults. A critical review and recommenda-
tions. Am J Prev Med 1998; 15: 316333.
24. Marshall AL, Owen N, Bauman AE. Mediated approaches for
influencing physical activity: update of the evidence on mass
media, print, telephone and website delivery of interventions. J Sci
Med Sport 2004; 7: 7480.
25. Napolitano MA, Marcus BH. Targeting and tailoring physical
activity information using print and information technologies.
Exerc Sport Sci Rev 2002; 30: 122128.
26. Orrow G, Kinmonth AL, Sanderson S, Sutton S. Effectiveness
of physical activity promotion based in primary care: systematic
review and meta-analysis of randomised controlled trials. BMJ
2012; 344: e1389.
27. van Sluijs EM, McMinn AM, Griffin SJ. Effectiveness of inter-
ventions to promote physical activity in children and adolescents:
systematic review of controlled trials. BMJ 2007; 335: 703.
28. Salmon J, Booth ML, Phongsavan P, Murphy N, Timperio A.
Promoting physical activity participation among children and ado-
lescents. Epidemiol Rev 2007; 29: 144159.
29. Dobbins M, De Corby K, Robeson P, Husson H, Tirilis D.
School-based physical activity programs for promoting physical
activity and fitness in children and adolescents aged 618.
Cochrane Database Syst Rev 2009; CD007651.
30. Cleland CL, Tully MA, Kee F, Cupples ME. The effectiveness
of physical activity interventions in socio-economically disadvan-
taged communities: a systematic review. Prev Med 2012; 54: 371
380.
2012 The Authors
obesity reviews 2012 International Association for the Study of Obesity
Activity interventions on low SES women V. Cleland et al. 15
31. Pan SY, Cameron C, Desmeules M, Morrison H, Craig CL,
Jiang X. Individual, social, environmental, and physical environ-
mental correlates with physical activity among Canadians: a cross-
sectional study. BMC Public Health 2009; 9: 21.
32. Liberati A, Altman DG, Tetzlaff J et al. The PRISMA state-
ment for reporting systematic reviews and meta-analyses of studies
that evaluate healthcare interventions: explanation and elabora-
tion. BMJ 2009; 339: b2700.
33. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred report-
ing items for systematic reviews and meta-analyses: the PRISMA
statement. BMJ 2009; 339: b2535.
34. Higgins J, Green S. Cochrane Handbook for Systematic
Reviews of Interventions Version 5.0.2 [updated September 2009].
URL http://www.cochrane-handbook.org/ (accessed October
2012).
35. Thomas BH, Ciliska D, Dobbins M, Micucci S. A process for
systematically reviewing the literature: providing the research evi-
dence for public health nursing interventions. Worldviews Evid
Based Nurs 2004; 1: 176184.
36. Chinn S. A simple method for converting an odds ratio to
effect size for use in meta-analysis. Stat Med 2000; 19: 31273131.
37. Higgins J, Green S. Cochrane Handbook for Systematic
Reviews of Interventions Version 5.1.0 [updated March 2011]. The
Cochrane Collaboration: URL http://www.cochrane-handbook.
org/ (accessed October 2012).
38. Abraham C, Michie S. A taxonomy of behavior change tech-
niques used in interventions. Health Psychol 2008; 27: 379387.
39. Alderson P, Green S. Additional module 1: meta-analysis of
continuous data. 2002.
40. Albright CL, Pruitt L, Castro C, Gonzalez A, Woo S, King AC.
Modifying physical activity in a multiethnic sample of low-income
women: one-year results from the IMPACT (Increasing Motivation
for Physical ACTivity) project. Anns Behav Med 2005; 30: 191
200.
41. Baranowski T, Simons-Morton B, Hooks P et al. A center-
based program for exercise change among Black-American fami-
lies. Health Educ Q 1990; 17: 179196.
42. Fahrenwald NL, Atwood JR, Walker SN, Johnson DR, Berg
K. A randomized pilot test of Moms on the Move: a physical
activity intervention for WIC mothers. Anns Behav Med 2004; 27:
8290.
43. Hovell MF, Mulvihill MM, Buono MJ et al. Culturally
tailored aerobic exercise intervention for low-income Latinas.
Am J Health Promot 2008; 22: 155163.
44. Jacobs AD, Ammerman AS, Ennett ST et al. Effects of a
tailored follow-up intervention on health behaviors, beliefs, and
attitudes. J Womens Health (Larchmt) 2004; 13: 557568.
45. Chang MW, Nitzke S, Brown R. Design and outcomes of a
Mothers In Motion behavioral intervention pilot study. J Nutr
Educ Behav 2010; 42: S11S21.
46. Olvera N, Bush JA, Sharma SV, Knox BB, Scherer RL, Butte
NF. BOUNCE: a community-based mother-daughter healthy life-
style intervention for low-income Latino families. Obesity (Silver
Spring) 2010; 18(Suppl. 1): S102S104.
47. Stoddard AM, Palombo R, Troped PJ, Sorensen G, Will JC.
Cardiovascular disease risk reduction: the Massachusetts WISE-
WOMAN project. J Womens Health (Larchmt) 2004; 13: 539
546.
48. Williams BR, Bezner J, Chesbro SB, Leavitt R. The effect of a
behavioral contract on adherence to a walking program in post-
menopausal African American women. Top Geriatr Rehabil 2005;
21: 332342.
49. Speck BJ, Hines-Martin V, Stetson BA, Looney SW. An
environmental intervention aimed at increasing physical activity
16 Activity interventions on low SES women V. Cleland et al.
levels in low-income women. J Cardiovasc Nurs 2007; 22: 263
271.
50. Lupton BS, Fonnebo V, Sogaard AJ; Finnmark Intervention
Study. The Finnmark Intervention Study: is it possible to change
CVD risk factors by community-based intervention in an Arctic
village in crisis? Scand J Public Health 2003; 31: 178186.
51. Lupton BS, Fonnebo V, Sogaard AJ, Langfeldt E. The
Finnmark Intervention Study. Better health for the fishery popula-
tion in an Arctic village in North Norway. Scand J Prim Health
Care 2002; 20: 213218.
52. Opdenacker J, Boen F, Vanden Auweele Y, De Bourdeaudhuij
I. Effectiveness of a lifestyle physical activity intervention in a
womans organization. J Womens Health (Larchmt) 2008; 17:
413421.
53. Wendel-Vos GCW, Dutman AE, Verschuren WMM et al. Life-
style factors of a five-year community-intervention program: the
Hartslag Limburg intervention. Am J Prev Med 2009; 37: 5056.
54. Brown WJ, Lee C, Oyomopito R. Effectiveness of a bilingual
heart health program for Greek-Australian women. Health
Promot Int 1996; 11: 117125.
55. Fjeldsoe BS, Miller YD, Marshall AL. MobileMums: a
randomized controlled trial of an SMS-based physical activity
intervention. Ann Behav Med 2010; 39: 101111.
56. Watson N, Milat AJ, Thomas M, Currie J. The feasibility and
effectiveness of pram walking groups for postpartum women in
Western Sydney. Health Promot J Austr 2005; 16: 9399.
57. Shirazi KK, Wallace LM, Niknami S et al. A home-based,
transtheoretical change model designed strength training interven-
tion to increase exercise to prevent osteoporosis in Iranian women
aged 4065 years: a randomized controlled trial. Health Educ Res
2007; 22: 305317.
58. Lucumi DI, Sarmiento OL, Forero R, Gomez LF, Espinosa G.
Community intervention to promote consumption of fruits and
vegetables, smoke-free homes, and physical activity among home
caregivers in Bogota, Colombia. Prev Chronic Dis 2006; 3: A120.
59. Wen CP, Wai JP, Tsai MK et al. Minimum amount of physical
activity for reduced mortality and extended life expectancy: a
prospective cohort study. Lancet 2011; 378: 12441253.
60. Isreal B, Schurman S. Social support, control, and the stress
process. In: Glanz K, Lewis F, Rimer B (eds). Health Behavior and
Health Education: Theory Research and Practice. Jossey-Bass Pub-
lishers: San Francisco, CA, 1990, pp. 187215.
61. Ball K, Abbott G, Cleland V et al. Resilience to obesity among
socioeconomically disadvantaged women: the READI study. Int J
Obes (Lond) 2012; 36: 855865.
62. Cleland V, Ball K, Hume C, Timperio A, King AC, Crawford
D. Individual, social and environmental correlates of physical
activity among women living in socioeconomically disadvantaged
neighbourhoods. Soc Sci Med 2010; 70: 20112018.
63. Cleland V, Ball K, King AC, Crawford D. Do the individual,
social and environmental correlates of physical activity differ
between urban and rural women? Environ Behav 2012; 44: 350
373.
64. Wolinsky FD, Stump TE, Clark DO. Antecedents and conse-
quences of physical activity and exercise among older adults. Ger-
ontologist 1995; 35: 451462.
65. Giles-Corti B, Donovan RJ. The relative influence of indi-
vidual, social and physical environment determinants of physical
activity. Soc Sci Med 2002; 54: 17931812.
obesity reviews
66. Courneya KS, McAuley E. Cognitive mediators of the social
influence-exercise adherence relationship: a test of the theory of
planned behavior. J Behav Med 1995; 18: 499515.
67. Ball K, Timperio A, Salmon J, Giles-Corti B, Roberts R, Craw-
ford D. Personal, social and environmental determinants of edu-
cational inequalities in walking: a multilevel study. J Epidemiol
Community Health 2007; 61: 108114.
68. Wilcox S, Castro C, King AC, Housemann R, Brownson RC.
Determinants of leisure time physical activity in rural compared
with urban older and ethnically diverse women in the United
States. J Epidemiol Community Health 2000; 54: 667672.
69. Eyler AA, Brownson RC, Donatelle RJ, King AC, Brown D,
Sallis JF. Physical activity social support and middle- and older-
aged minority women: results from a US survey. Soc Sci Med 1999;
49: 781789.
70. Cleland V, Ball K. What might work? Exploring the perceived
feasibility of strategies to promote physical activity among women
living in socioeconomically disadvantaged neighbourhoods.
Health Educ Res 2012; doi:10.1093/her/cys097.
71. Trost SG, Owen N, Bauman AE, Sallis JF, Brown W. Corre-
lates of adults participation in physical activity: review and
update. Med Sci Sports Exerc 2002; 34: 19962001.
72. Welch N, McNaughton SA, Hunter W, Hume C, Crawford D.
Is the perception of time pressure a barrier to healthy eating and
physical activity among women? Public Health Nutr 2009; 12:
888895.
73. Withall J, Jago R, Fox KR. Why some do but most dont.
Barriers and enablers to engaging low-income groups in physical
activity programmes: a mixed methods study. BMC Public Health
2011; 11: 507.
74. Andajani-Sutjahjo S, Ball K, Warren N, Inglis V, Crawford D.
Perceived personal, social and environmental barriers to weight
maintenance among young women: a community survey. Int J
Behav Nutr Phys Act 2004; 1: 15.
75. Baranowski T, Cullen KW, Baranowski J. Psychosocial corre-
lates of dietary intake: advancing dietary intervention. Annu Rev
Nutr 1999; 19: 1740.
76. Cliska D, Miles E, OBrien MA et al. Effectiveness of
community-based interventions to increase fruit and vegetable
consumption. J Nutr Educ 2000; 32: 241252.
77. Glanz K, Bishop DB. The role of behavioral science theory in
development and implementation of public health interventions.
Annu Rev Public Health 2010; 31: 399418.
Supporting Information
Additional Supporting Information may be found in the
online version of this article:
Appendix S1. PRISMA checklist for a systematic review or
meta-analysis of randomized trials and other evaluation
studies.
Appendix S2. Search strategy and results.
Appendix S3. Data extraction.
Appendix S4. Behaviour change techniques employed in
studies, categorized according to the Abraham taxonomy.
Appendix S5. References for supporting information.
2012 The Authors
obesity reviews 2012 International Association for the Study of Obesity