If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Mannheim Institute of Public Health, Social and Preventive Medicine, Medical Faculty Mannheim, Heidelberg University, GermanyDepartment of Medicine, Case Comprehensive Cancer Centre, Case Western Reserve University, Cleveland, OH, USADepartment of Epidemiology and Biostatistics, Case Comprehensive Cancer Centre, Case Western Reserve University, Cleveland, OH, USA
Despite the known benefits of physical activity, the majority of adults in developed countries lead sedentary lifestyles. The community setting is a promising venue for physical activity-promoting interventions. Our objectives were to investigate the effectiveness of community-based physical activity interventions by mode of delivery, study quality and to analyse intervention effectiveness in different subgroups in the population.
We conducted a systematic literature review in Medline and other databases to identify controlled, community-based physical activity interventions published between 2001 and 2012.
We performed several post hoc subgroup comparisons for mode of delivery, study quality and selected population characteristics, using net per cent change in physical activity outcomes between baseline and follow-up as an effect measure.
We identified 55 studies on exercise/walking sessions, face-to-face counselling, public campaigns and interventions by mail, the Internet and telephone presenting data on 20,532 participants. Overall, half of the studies reported positive physical activity outcomes (total net per cent change: 16.4%; p = 0.159; net per cent change for high-quality studies, i.e. studies meeting more than 5 out of 7 quality criteria: 16.2%; p = 0.010). Interventions using face-to-face counselling or group sessions were most effective (net per cent change: 35.0%; p = 0.014). Net per cent change was also higher in studies exclusively tailored to women (27.7%; p = 0.005) or specific ethnic groups (38.9%; p = 0.034).
This systematic review supports the effectiveness of community-based physical activity interventions in high-quality studies. Our results suggest that interventions using personal contact as well as tailored interventions are most promising.
Regular physical activity (PA) is a key factor in the prevention and treatment of many chronic diseases. It is associated with increased physical and psychological well-being and reduced risk of all-cause mortality.
Acknowledging the health benefits of PA in the general population, it is recommended that all healthy adults engage in moderate-intensity PA for a minimum of 30 min on five days per week or in vigorous PA for a minimum of 20 min on three days per week.
Given significant gaps between recommendations and self-reported levels of activity, increasing the population level of PA has become a leading area of focus in contemporary public health policy. Earlier reviews have demonstrated the effectiveness of community-based PA interventions using face-to-face interaction in small groups, mail or telephone contact and community-wide campaigns in increasing PA.
Since the beginning of the new century, newer approaches using technologies like e-mail and the Internet have been used increasingly to disseminate public health information. Previous studies suggest the promise of incorporating new technology in the delivery of PA interventions.
However, traditional approaches for promoting PA may still be more appropriate in some community settings. Previous work documents, for example, that Internet use among subgroups at higher risk for cardiovascular morbidity and mortality (e.g., older or less educated persons and those living in rural areas) is lower.
To date, little is known about the comparative effectiveness of community-based PA interventions using different modes of delivery. Additionally, there is a relative lack of data on the effectiveness of such interventions in deprived subgroups of the population in whom PA interventions may prove most valuable. Furthermore, previous work suggests significant variability in study quality and this may affect the conclusions that can be drawn. In this paper, we provide an overview on the effectiveness of community-based PA interventions through a systematic review of recent literature. Our objectives were to investigate the effectiveness of PA interventions by mode of delivery, study quality and in different subgroups in the population.
The following computerized databases were searched for original research articles published between January 1, 2001 and June 30, 2012: Medline, PsycINFO, CSA Sociological Abstracts and SPOLIT. The following keywords and search strategy were chosen using the Medical Subject Headings thesaurus: (physical activity OR exercise) AND (randomized controlled trial OR intervention OR health promotion) AND (community OR community setting). The initial screen produced n = 2302 hits from the four databases. We further examined the reference lists of articles in an effort to identify all relevant publications. Following the initial application of exclusion criteria (as described below) to information contained in the study abstract, the number of hits was reduced to 80. The pool of potentially eligible studies was reduced further to n = 55 following review of the entire manuscript (Fig. 1). Wherever possible, the procedure used in this study follows the PRISMA statement, which was developed to guide the reporting of systematic reviews.
Approval by an ethics committee was not necessary because only published data were used. The authors followed the principles outlined in the Declaration of Helsinki and the guidelines on good epidemiological practice.
Studies from developed countries published in English were included in the review if they contained the following elements: (a) adults aged 18+; (b) PA intervention initiated in a community setting; (c) randomized controlled trial (RCT) or quasi-experimental study with a comparison group; and (d) reported outcomes including at least one measure of PA. We defined “community” as an administrative or geographical boundary area (place of residence) or as social networks that were generally open to a large part of the population (e.g., churches). In contrast, we excluded interventions conducted in clinical or occupational settings (e.g., health care or workplace) because they comprise a separate body of literature. Furthermore, studies focusing on clinically-defined subgroups of the population (e.g., obese individuals or those with a specific clinical diagnosis), were also excluded as they are not representative of the general population in communities. We focused on studies with both an explicit intervention to increase PA and a measure of post-intervention PA behaviour reported with sufficient detail to calculate effect estimates. Only studies from developed countries were considered in this review given concerns for substantial differences that may exist in lifestyles, social structures and the “built” environment that might contribute to different opportunities for PA. Methodologically, such variability from these and other sources would increase heterogeneity and non-comparability within the sample of reports under consideration. We also excluded studies in which the sole outcome measures were motivation to exercise or self-efficacy, as these may not necessarily translate into action. Other exclusion criteria are shown in Fig. 1.
Each study included in this review was evaluated using a standardized abstraction form. We specifically assessed the mode of intervention delivery, length of intervention and follow-up, treatment in experimental and control groups and the theoretical basis of the intervention. Outcome measures of PA and other health effects (e.g., improvement in physiologic parameters, other behaviour or knowledge) as well as level of significance were extracted if reported. Wherever possible, we extracted data on the proportion of participants achieving a sufficient level of PA as defined by ACSM/AHA recommendations,
to enable cross-study comparisons. If these data were unavailable, we extracted data on the total minutes of moderate to vigorous PA, total steps per week or scores based on metabolic equivalents (MET). The mode of delivery was classified as face-to-face counselling/group sessions, exercise/walking sessions, mail- or telephone-mediated interventions, public campaigns or studies using e-mail, computer- or web-based formats for intervention delivery. Studies applying several modes of delivery were classified as multicomponent in nature.
We judged the quality of results in each study using a previously described approach based on the extent to which seven binary criteria were met.
These included: randomization, exclusion of exposure contamination in the control group, representativeness of the sample, comparability of intervention and control group, attrition rate <30% or sample size > 100 in each group, sufficient period for PA data collection and use of a valid instrument for PA assessment. For the studies in the analytic sample, two authors (CB, MNJ) independently determined whether each criterion was fulfilled. Per cent agreement for these criteria was good and ranged between 0.6 and 1.0. Finally, values for each criterion were summed to form a quality score.
Net per cent change (NPC) in PA was calculated using the formula by Kahn et al.
We determined the net per cent change in PA from baseline to follow-up in the intervention (I) and the control (C) groups as [((Ipost − Ipre)/Ipre) − ((Cpost − Cpre)/Cpre)] × 100%. In subgroup analyses, we assessed differences in NPC by mode of intervention delivery, study quality and selected population and study characteristics such as age (e.g., older adults defined as mean age 50+), sex, ethnicity (percentage Caucasian), socioeconomic status (SES), region (North America, Europe, Asia, Australia/New Zealand), short- (≤6 months) and long-term (>6 months) follow-up. To assess intervention effectiveness by a specific study condition, we used individual one-sample t-tests to determine if the average NPC in these studies weighted by sample size significantly differed from zero. Because we encountered substantial heterogeneity, we were unable to conduct a meta-analysis and instead present NPC with unadjusted 95% confidence intervals (95%-CI). All statistical analyses were conducted with SAS 9.2 (SAS Institute Inc., Cary, USA) with a two-sided alpha level of p < 0.05.
Our sample was comprised of 37 RCTs and 18 quasi-experimental studies. More than half of these studies (n = 31) were conducted in the U.S., twelve in Australia/New Zealand, ten in Europe and two in Asia. The median total quality score was 5 (range: 3–7). Forty-three studies had sufficient sample sizes and almost all studies measured PA over at least a one-week period and used previously validated measures for outcome assessment. Only a few studies included representative samples of the general community population; 23 studies focused on previously underactive adults and 20 reported on PA in samples restricted to women only. The total number of participants across the 55 studies was 20,532; individual sample sizes ranged from 31 to 3114 (median: 154). Across studies, participants had a weighted mean age of 50.1 years, were predominantly female (66.9% [95%-CI: 62.3%; 71.6%]), married (64.3% [58.6%; 70.0%]) and Caucasian/white (64.5% [50.4%; 78.6%]). A summary of the study characteristics and quality score is provided in the supplemental Table S2.
There was substantial variation in the modes used to recruit participants for PA interventions including advertisements, distribution of flyers, local media, word-of-mouth as well as recruitment in churches and community centres. Most studies (n = 50) used traditional modes of intervention delivery, while five studies used newer modes like e-mail, computer- or web-based formats. Among the studies with traditional modes of intervention delivery, there were ten studies with exercise or walking sessions including wearing pedometers/accelerometers, eight studies with face-to-face counselling or group sessions, six public campaigns, five mail- and two telephone-mediated interventions. Nineteen studies used more than one approach for intervention delivery and were classified as multicomponent interventions.
Intervention duration ranged from a single contact to up to five years (median: 6 months). In the majority of studies, follow-up of participants was completed by the end of the intervention (median length of follow-up after end of intervention: 0 weeks; range: 0 weeks to 3 years). Most studies used questionnaires to assess outcome measures of interest (e.g., weekly amount (n = 31) or proportion of sufficiently active persons (n = 16)) and eight studies provided data on step counts derived from pedometers or accelerometers. Seventeen out of the 55 studies included multiple lifestyle interventions (e.g., PA and dietary intervention). PA interventions were tested in most cases against no treatment or a waiting list (n = 34). Comparison groups in other studies received treatment other than PA or minimal treatment with parts of or similar PA elements as the intervention group. Most studies were theory-based (n = 46), incorporating some aspect of Social Cognitive Theory, the Transtheoretical Model or other models of behaviour change into their design.
A qualitative description of the intervention effects in individual studies is presented in supplemental Table S1. NPC values for all studies are given in the supplemental Table S2. Positive changes in PA behaviour were reported in 25 studies, 16 studies showed no significant effect of the intervention on PA behaviour and in the remaining 14 studies the effect of the intervention was unclear. The grand mean difference in the NPC in PA outcomes weighted by study sample size was 16.4% [−6.6%; 39.5%] (p = 0.159). NPC values differed significantly from zero only when studies with quality scores above the median were considered (NPC = 16.2% [4.4%; 28.0%]; p = 0.010).
There were several characteristics of intervention or study design also associated with significant NPC values. Results from these subgroup analyses are given in Fig. 2. For two modes of delivery, mean NPC values differed significantly from zero: face-to-face counselling/group sessions (NPC = 35.0% [9.6%; 60.5%]; p = 0.014) and mail-mediated interventions (NPC = 18.9% [2.2%; 35.6%]; p = 0.035). NPC values for all other modes are given in Fig. 2. Although mean NPC was higher for multi- vs. single component interventions, this difference did not achieve statistical significance (35.9% vs. 9.9%; p for between-group difference = 0.505). Intervention effectiveness was significant in studies with intervention duration of more than twelve months or in those with sample sizes greater than 150 participants (both p < 0.05). We also observed that mean NPC values were greater in studies reporting short-term follow-up (p = 0.011), but not in those reporting long-term follow-up (p = 0.569). There was no difference in mean NPC by study region.
Compared with all other studies, mean NPC was greater in those that focused exclusively on women (NPC = 27.2% [9.3%; 46.1%]; p = 0.005). Of these, two interventions specifically addressed mothers with young children
Twenty-two studies included middle-aged or older adults and had a mean NPC of 24.0% [−35.9%; 83.8%] (p = 0.414). Effective interventions in older adult populations included those using more modern approaches – web-based interventions, computer-controlled interactive telephone systems and hand-held computers.
Additionally, nine studies limiting recruitment to non-Caucasian populations and providing interventions culturally tailored to ethnic subgroups reported significantly positive mean NPC values (38.9% [3.7%; 74.2%]; p = 0.034). In contrast, four of eight studies focusing on persons with low SES reported positive PA outcomes, although the overall mean NPC was low (NPC = 7.7% [−6.7%; 22.0%]; p = 0.248; Fig. 2). Further analysis of the effectiveness of PA interventions in different social groups was not possible due to limited information provided in the original research articles.
This systematic review supports the effectiveness of community-based PA interventions in high-quality studies, with positive outcomes in approximately half of the studies in our sample. Comparisons by mode of intervention delivery suggest that interventions delivered through personal contact as well as tailored interventions appeared most effective and that newer approaches appeared to add relatively little to intervention effectiveness compared to more traditional ones. The overall effectiveness of community-based PA-promoting interventions differed by sex, ethnicity and characteristics of study design.
Our findings share similarities with previous studies on community-based PA interventions. The earlier review conducted by Kahn and colleagues, for example, reported a median net increase of 4.2% in the proportion of physically active persons for traditional, community-wide interventions.
That interventions using newer modes of delivery were not superior to more traditional approaches might be due to less personal support, less tailoring or technological barriers related to this type of delivery. On the other hand, studies using newer modes of intervention delivery are relatively under-represented in the current review, affecting power to detect significant differences in effect across different approaches.
Targeted interventions within community settings seem to be effective in increasing short-term effects, an observation supported in previous work.
the availability of only a few studies with long-term follow-up limits clear conclusions on sustained effectiveness. Other studies included in the present review suggested that interventions can be effective also in selected ethnic groups particularly if they are culturally tailored. However, information on ethnic minorities or those with low SES was generally too limited to reach a clear conclusion on the effectiveness of PA interventions in these important subgroups, often at higher risk for cardiovascular morbidity and mortality.
for example, successfully recruited participants from the general population using project-staff-involved recruitment strategies (80%), while mass mailings, posters and media campaigns contributed relatively little (16%) to the total number of recruited participants. Other techniques involving some form of personal contact including word-of-mouth communication appear useful in promoting awareness of PA programs.
The current review is consistent with previous work insofar as it points to the generally greater effectiveness of interventions that are culturally tailored or that involve some degree of personal contact. However, it is important to acknowledge that conclusions on the reach or overall impact of PA interventions individually or by subgroups cannot be drawn given the limited information found in the articles we identified for review.
This report updates a previous review of evidence for the effectiveness of PA interventions in community settings by modes of delivery, study quality and selected population characteristics. Original studies were identified from several databases enabling the evaluation of a range of potential approaches to promote PA based in a far-ranging group of disciplinary perspectives. Although some studies included in our review failed to meet all of the quality criteria we applied, a focus on RCTs and quasi-experimental studies with control groups reduces several potential sources of bias that can affect quality. We accounted for heterogeneity between studies arising from different outcome metrics by expressing effect size using NPC that did not require the presence of a comparable outcome across studies and was weighted by sample size.
This study is subject to several limitations. First, the number of studies included in our subgroup analyses was rather low, raising the possibility of limited power in identifying significant differences in the effectiveness by mode of PA intervention. Furthermore, heterogeneity was present between studies that arose from different types of PA intervention, the specific subgroups under study and differing durations of follow-up. Another limitation is that most studies relied on self-reported PA, introducing the possibility that social desirability bias may have been present. Additionally, results based on interventions tailored to specific cultural backgrounds or low-income populations may not be generalizable to other groups or settings. Conversely, it is unclear whether interventions effective in the general population will be similarly successful in specific subgroups.
Community-based interventions appear generally effective in promoting PA. While the community setting may be suitable for reaching a large population, diversity within communities, however, may make it difficult and potentially undesirable to recruit people with different characteristics into a single “one size fits all” intervention program. Tailored recruitment and interventions may therefore prove useful in acknowledging differences in preferences for programs to promote PA. Future studies may consider incorporating measures of intervention reach and other dimensions of program evaluation as comprehensively as intervention effectiveness to generate estimates for overall impact of PA interventions.
Community-based PA interventions appear most effective when they include some form of personal contact for intervention delivery (e.g., face-to-face counselling or group sessions).
Tailoring is important not only for intervention content or delivery, but also for the recruitment of participants.
Future studies should have a proper design and incorporate measures of intervention reach and other dimensions of program evaluation.
There has been no financial assistance with this project. We thank Katharina Enz (Mannheim Institute of Public Health, Germany) for her assistance in obtaining the literature.