Interventions to promote or maintain physical activity during and after the transition to retirement: an evidence synthesis

The population (patient) group

This systematic review considered older adults who were about to retire or who had recently retired.

The intervention

Interventions examined had the reported aim of maintaining or increasing physical activity, delivered in any context and by any method.

Comparator

The review included studies with any comparator or no comparator.

Outcomes

Outcomes of interest were those relating to physical activity levels recorded by validated measures/scales, or other indirect measures such as hours of activity, well-being or measures of physical or mental health.

Search strategies

A systematic and comprehensive literature search of key health and medical databases was undertaken from March to December 2014. The searching process aimed to identify studies that reported the effectiveness of interventions for people around the transition to retirement, as well as studies that reported the views and perceptions of older adults and staff regarding interventions. Searching was carried out for both reviews in parallel, with allocation to either effectiveness or qualitative reviews at the point of identification and selection of studies for potential inclusion. The search process was recorded in detail and included lists of databases searched, the dates on which searches were run, the limits applied, the number of hits and duplication as per Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. 15 The search strategy is provided in Appendix 1.

The search strategy was developed by the information specialist on the team (Helen Buckley-Woods) who undertook electronic searching using iterative methods to create a database of citations using Reference Manager version 12 (Thomson ResearchSoft, San Francisco, CA, USA). The searching process was both iterative and emergent,16 with the results of an initial search informing the development of a subsequent search strategy (and so on), in order to explore fully the evidence relevant to the research questions. Database searching alone is not sufficient to provide the necessary evidence for a systematic review17,18 so ‘supplementary’ searching techniques, such as citation searching, were employed to ensure a comprehensive approach to identifying the evidence. As public health evidence is typically dispersed across a number of disciplinary fields, a wide variety of data sources were searched to include medicine and health, social science and specialist bibliographic databases (such as SPORTDiscus). Sources to identify grey literature were also explored. Searches were limited to 1990 to present where sources allowed.

First search

An initial search was developed using terms that reflect the concept of the transition into retirement combined with terms that reflect the concept of physical activity. ‘Retirement’ terms were broad to reflect the population of interest and the varied circumstances of ‘retirement’ (i.e. not only from full-time employment). Hence, terms for redundancy, ‘empty-nest’ and other similar circumstances of role change (such as becoming a mature student) were included. The search strategy was informed by an existing scoping search and incorporated suggestions for terms and concepts from the project team and as a result of consultation in stakeholder workshops.

Second search

The initial search retrieved a limited number of papers that specifically referred to the transition to retirement or to retirement as a concept. In order to assess whether or not this was a true reflection of the published literature, a new search was developed which used terms for older age to reflect the population of interest. These terms were combined with the terms for physical activity that were used in the initial search. Combining only these two facets of the question would retrieve unmanageable numbers of papers, so a modified study filter was applied to identify quantitative studies of either clinical trial or observational design.

Citation searching

In addition to standard electronic database searching, citation searching was undertaken later in the project (November–December 2014). Two sets of papers were used for citation searches. First, citations to a small set of existing reviews on the topic of interest were retrieved. Second, citations to both qualitative and quantitative included papers were identified, producing a further substantial set of results.

Grey literature

Searches for grey literature were undertaken in order to identify any reports or evaluations of ‘grass roots’ projects or other evidence not indexed in bibliographic databases and also to minimise problems of publication bias. These searches were either in specific grey literature sources such as ‘Index to Theses’ and ‘OpenGrey’ or by searching specific topic-relevant websites.

Sources searched

The following electronic databases were searched for published and unpublished research evidence from 1990 onwards:

• Applied Social Sciences Index and Abstracts (via ProQuest).

• Cumulative Index to Nursing and Allied Health Literature (via EBSCOhost).

• Cochrane Library, including Database of Abstracts of Reviews of Effects, Cochrane Central Register of Controlled Trials, Health Technology Assessments and Cochrane Database of Systematic Reviews (via Wiley).

• Health Management Information Consortium (via Ovid).

• MEDLINE and MEDLINE In-Process & Other Non-Indexed Citations (via Ovid).

• Science Citation Index via Web of Knowledge (Thomson ISI).

• Social Science Citation Index via Web of Knowledge (Thomson ISI).

• Sociological Abstracts (via ProQuest).

• PsycINFO (via Ovid).

• Social Policy and Practice (via Ovid).

• SPORTDiscus (via EBSCOhost).

• Evidence for Policy and Practice Information and Co-ordinating Centre Databases: Bibliomap, Database of Promoting Health Effectiveness Reviews, Trials Register of Promoting Health Interventions.

• The database on Obesity and Sedentary behaviour studies: http://eppi.ioe.ac.uk/.

• Open Grey: www.opengrey.eu/.

• Index to Theses: www.theses.com.eresources.shef.ac.uk/default.asp.

• Conference Papers Index Science and Social Science (via Web of Science).

• Department of Health: www.dh.gov.uk.

• BHF national centre for physical activity: www.bhfactive.org.uk/.

• U3 A: www.u3a.org.uk/.

• Web search (Duck Duck Go): https://duckduckgo.com/.

All citations were imported into Reference Manager (version 12) and duplicates deleted prior to scrutiny by members of the team.

Search restrictions

Searches were limited by date (1990 to present), as the review was aiming to synthesise the most up-to-date evidence. The searches did not set an English-language restriction. We included studies published in developed countries and, although we intended that the review would be predominantly limited to work published in English to ensure that papers were relevant to the UK context, we aimed to search for and include any additional key international papers that had an English abstract and that we were able to obtain.

Inclusion and exclusion criteria

Effectiveness studies

Data were synthesised in a form appropriate to the data type. It was proposed that meta-analysis calculating summary statistics would be used if heterogeneity permitted, with use of graphs, frequency distributions and forest plots. It was anticipated that subgroups including age of participants, learning disability, intervention content and delivery agent would be examined if numbers permitted. The heterogeneity of the included work precluded summarising the studies via meta-analysis, as described in Chapter 3, Intervention effectiveness by typology. In order to provide a visual summary, we drew on Harvest plot techniques to examine intervention outcomes across the different types and to perform comparison of where there was greater evidence of rigour. 23

Effectiveness review findings were reported using narrative synthesis methods. We tabulated characteristics of the included studies and examined outcomes by characteristics such as intervention content, agent of delivery, intervention dosage and length of follow-up. Relationships between studies and outcomes within these typologies were scrutinised.

Qualitative studies

Qualitative data were synthesised using thematic synthesis methods in order to develop an overview of recurring perceptions of potential obstacles to successful outcomes within the data. 24 This method comprises familiarisation with each paper and coding of the finding sections (which constitute the ‘data’ for the synthesis) according to key concepts within the findings. Although some data may directly address the research question, sometimes information such as barriers and facilitators to implementation has to be inferred from the findings, as the original study may not have been designed to have the same focus as the review question. 24

Combining quantitative and qualitative data

We had intended to use a meta-synthesis method which uses the qualitative data to add explanatory value to effectiveness study findings. 25,26 However, we found little of the qualitative literature referred specifically to interventions; instead, it tended to report more general views and perceptions of physical activity in retirement and/or older age, as well as elements of activity that may be most appealing. We therefore used a meta-synthesis method whereby themes relating to facilitators of engagement in physical activity were used to examine the content of interventions included in the review of effectiveness.

Study design

Figure 3 shows that the identified literature was of a reasonably high quality in terms of study design, with a large proportion (36) of the papers reporting studies using a randomised controlled design,28–32,34–36,39,41,44,46–50,52,53,55–59,61,62,69,71,73–78,83,84,86 and 18 papers reporting studies using a cluster randomised design. 27,37,38,40,42,43,45,54,66–68,70,79–82,85 See Appendix 3 for lists of the studies within each category.

FIGURE 3.

Number of intervention studies of each design. CBA, controlled before and after.

Country of origin

An overview of included studies by country of origin is provided in Figure 4. The greatest proportion of work was reported by authors based in the USA (33 papers),27,30,32,34,36,39,41–43,45,46,50–53,55–57,60–62,69,71–73,75,77,83,84 followed by authors based in the Netherlands and in Australia/New Zealand. Three of the papers were from the UK. 33,74,76 (See Appendix 3 for the list of studies within each category.)

FIGURE 4.

Number of intervention studies from each country.

Sex

The literature contained 34 papers with either all female participants or a majority of female participants28,30,31,34–36,40,41,46,48–53,55–59,61,62,69,72–75,77,83,84,87–90 (Figure 5) (see Appendix 3 for a list of studies). This contrasted with only two studies that recruited only males. 27,85

FIGURE 5.

Number of intervention studies by sex of participants.

Socioeconomic/educational status

We identified only one study that described participants as being of predominantly low SES29 and one paper that described participants from a minority ethnic population. 75 Sixteen papers described their studies as being in predominantly more highly educated/higher income participants. 30,32,34–36,41,46,50,52,53,61,62,65,69,73,77 The majority of studies were either unclear regarding education/SES or included diverse/mixed participants. See Figure 6 for a summary graph and Appendix 3 for lists of studies within each category.

FIGURE 6.

Number of studies reporting socioeconomic characteristics of participants.

Physical activity level

Approximately half of the studies (34) recruited participants who were below recommended activity levels, with the majority of these using higher activity level as an exclusion criterion. 27–31,34–36,40,41,46,50,52,53,55–57,59–65,70,71,73,75,77,86–90 The other 29 papers32,33,37–39,42–45,47,48,51,54,58,66–69,72,74,76,78–85 did not detail inactivity as being an inclusion criterion, and therefore in the absence of any other reporting it is assumed that these participants were most likely to be of mixed activity levels. (See Appendix 3 for detail of the studies within these categories.)

Age of participants

As outlined above, in the absence of literature specifically examining people at retirement transition, we took the decision to grade the applicability of other papers that reported interventions in older adults, with the average age of participants standing as a proxy for retirement age. We developed a four-point applicability rating system, in which A1 studies were those that referred to participants as immediately before or after retirement, A2 studies were those with an average participant age in the most common retirement transition age range of 55–69 years, and A3 studies were those that were less likely to be applicable to the retirement transition period, having an average age of either 50–54 years (A3i) or 70–75 years (A3ii).

This rating system was based on the most common statutory retirement ages across OECD countries of 65 or 67 years. The band 55–69 years was, therefore, most likely to include those approaching retirement (and those taking early retirement) together with those in the period immediately following retirement, and was of most relevance to our research question. The A3 papers, although potentially relevant, were considered to be of lower applicability. These papers were identified and examined to ascertain whether or not there were any studies of particular significance to the review. Table 4 presents the papers rated A1 and A2, categorised by age range. In this report, we intend to focus our synthesis on these A1/2 studies, as those most relevant to the retirement transition. However, we will provide an overview of the A3 studies and also consider any similarities and differences between the A1/A2 group and A3 studies.

Retirement/employment

As described above, only a minority of papers made reference to the employment/retirement status of the older adults included in the studies. The 22 papers that provide this detail are listed in Table 5. As can be seen, four of the interventions were carried out in populations in which all individuals were described as retired. Where percentages of those employed are provided, it is not clear if the other non-employed participants were unemployed or retired or working part time. The challenge in identifying the retirement status of study populations in these papers was further highlighted by Hooker et al. 51 This study makes an interesting distinction between those participants that are categorised as ‘retired and working’ and those categorised as ‘retired and not working’. A further point of interest relates to the only paper we identified to describe the study population as recently retired. This paper85 gives the mean age of study participants as a seemingly young age of 59.5 years, which supports our decision to include those in their late fifties in our A2 applicability band.

Intervention typology

The papers reported a varied range of intervention approaches, and included one study that intervened with health-care professionals in order to enhance the content of consultations,57 one that evaluated a community campaign54 and a further study that analysed the effect of providing free bus passes to older adults. 33 The remaining papers were divided into those that evaluated interventions provided to participants within their home (content was delivered via the telephone, via e-mail/internet or via post) and those where participants either travelled to their local community health centre/surgery for advice/counselling, attended classes/workshops, or took part in organised walks/swimming sessions. For many of these away-from-home interventions, it was unclear where exactly the sessions were held.

The in-home interventions often included multiple elements such as advice, pedometer use, keeping an exercise diary, and/or information. A number of the papers examined different variants of interventions rather than comparing with a control group. The Peels et al. 66–68 studies, for example, varied the delivery method (via the internet vs. via mail), and the content (additional environmental information vs. none). Croteau et al. 36 examined the addition of group sessions to a standard intervention, Sawchuk et al. 75 evaluated the addition of a pedometer to their programme and Strath et al. 77 included telephone calls for a subgroup of participants.

Outcomes measured

The included literature measured a wide range of outcomes, encompassing those that were self-reported via completion of questionnaires, in person, via telephone or via postal questionnaire. Outcomes that were measured by the research team included weight, body mass index (BMI) and fitness tests, together with data downloaded from pedometers or accelerometers. Table 6 outlines the physical activity measures that were used within the set of papers. Of note are only two papers which measured inactivity in addition to activity. 29,55 The first of these29 collected data on reported sitting time (minutes per week), and the second55 used a questionnaire (Measure of Older Adults’ Sedentary Time), with television viewing time considered to be the primary outcome of interest. In addition to these measures relating to physical activity, many papers also included a raft of other measures such as strength, balance, flexibility or falls, which we have not included as not directly relating to activity levels.

Follow-up periods

Whereas 10 studies carried out outcome assessment immediately following the intervention,27,29,34,41,50,57,58,73,75,77 nine studies reported follow-up periods of > 12 months30,32,43,59,61,63,64,69,78,85 (Table 7).

Education of health-care professionals

One paper57 evaluated an intervention with staff which trained primary care providers to offer referrals to community exercise programmes for patients who reported before their clinic visit that they were ‘contemplative’ about regular exercise. This cluster RCT study found that patients were more likely to receive exercise advice from trained GPs. However, there was no significant difference in patient-reported regular exercise for the intervention group compared with controls.

Counselling and advice

Eleven included papers28,34,36,40,45,47,59,60,70,71,76 (10 studies) assessed the effectiveness of interventions comprising the giving of advice or counselling with seven of these providing stronger evidence of effect. One of these interventions was delivered by peer mentors, one by trained physicians, one by a nurse, two by exercise professionals, two encouraged patients to prompt their physician, and the final four papers examined combined physician and exercise professional input.

The Stevens et al. 76 paper from the UK evaluated both clinical effectiveness and cost-effectiveness of a 10-week programme, in which inactive people were invited to a consultation with an exercise professional. They were given information on physical activity and health and the local opportunities available to them to be more active. This study had a participant mean age of 59 years and found a net 10.6% [95% confidence interval (CI) 4.5% to 16.9%] reduction in the proportion of people classified as sedentary in the intervention group compared with the control group. In addition, the study found an increase in the mean number of self-reported episodes of physical activity per week in the intervention group compared with the control group (an additional 1.52 episodes, 95% CI 1.14 to 1.95 episodes) at 8 months’ follow-up. The cost of achieving the recommended level of activity for each person was estimated at £2500.

A study evaluating physician exercise prescriptions in Canada70 had two intervention arms, with prescription only and prescription plus counselling programmes. Around half the participants in this cluster RCT study were retired, with all having an inactive lifestyle. Cardiorespiratory fitness increased significantly (p < 0.001) at 12 months’ follow-up for both groups [prescription + counselling 3.02 ml of oxygen per kilogram of body weight per minute (ml/kg/minute) (95% CI 2.40 to 3.65 ml/kg/minute), and prescription only 2.21 ml/kg/minute (95% CI 1.27 to 3.15 ml/kg/minute) with no significant difference between the groups]. Physical activity was measured by 7-day recall, with estimated kilocalories/kilograms per day significantly increasing in both groups. However, the prescription plus counselling group increased significantly more than the prescription-only group (p = 0.006). A sex difference in response was noted, with women responding significantly better (improved predicted maximal oxygen consumption p < 0.001) to the prescription plus counselling intervention than prescription-only, and men benefiting equally from the interventions.

Lawton et al. 59 examined outcomes following a brief advice session led by a nurse for women with low activity levels (mean age 59 years). The counselling session was supplemented by monthly telephone support and a follow-up session after 6 months. Mean self-reported physical activity levels were higher for intervention participants than for controls (p < 0.001) and a greater proportion in the intervention group than the control group reached the target of physical activity at 12 months [233 (43%) vs. 165 (30%); p < 0.001]. Levels declined but were still significantly different at 2 years [214 (39%) vs. 179 (33%); p < 0.001]. There was no difference between groups with regard to clinical outcomes and it was noted that falls/injuries were higher in the intervention groups.

The third study providing stronger evidence of effectiveness36 had two intervention arms, with one consisting of counselling, pedometer use and self-monitoring over 6 months, and the other consisting of the same interventions plus monthly mentoring meetings with a fitness professional. Participants in this small-scale evaluation (described as a pilot study) were inactive, with a mean age of 64 years, and 16 of 36 participants were employed. The study found a significant intervention effect [p = 0.015; effect size (ES) 0.611] with both groups significantly increasing step count at 6 months, with no significant difference between groups (p = 0.151). The group that received additional mentoring maintained the effect at 12 months (ES 0.606), whereas there was fading of effect for the standard intervention group.

Another paper comparing two interventions71 evaluated advice from a physician and physician advice supplemented by telephone counselling from a health educator, for patients with a mean age of 68 years. At 3 months, 7-day recall self-reported data indicated significantly increased minutes of moderate physical activity for the advice plus counselling group (p < 0.05), increased moderate exercise kilocalories per week (p < 0.05) and biotrainer physical activity counts (p < 0.01) for this group from baseline. These improvements were maintained at 6 months. The physician advice-only group also increased on most measures, although the changes were not statistically significant.

In a small-scale pilot study, Armit et al. 28 randomised participants to three intervention arms: advice from a GP; counselling and telephone calls from an exercise scientist; or counselling and telephone calls with the addition of pedometer use. The participants were inactive patients recruited from physician waiting rooms with a mean age of 64 years. At 12 weeks, there was an overall increase of 116 weighted minutes of self-reported physical activity per week (p < 0.001) across the three groups. The paper describes no significant difference between groups, although the data presented indicate no change in the GP advice group compared with significant differences in average physical activity levels pre–post for the other two groups.

Another similar intervention with GP advice, supplemented by telephone support from an exercise specialist,40 increased self-reported mean total energy expenditure by 9.4 kcal/kg/week (p = 0.001) and leisure exercise by 2.7 kcal/kg/week (p = 0.02). This equated to 34 minutes per week more in the intervention group than in the control group (p = 0.04). The proportion of the intervention group reporting undertaking 2.5 hours per week of leisure exercise also increased by 9.72% (p = 0.003) more than in the control group. The participants were sedentary patients visiting their GP/nurse (mean age 57 years), and the intervention was designed to prompt practitioners to give advice and an exercise programme to patients on being given a form by patients.

A paper from Australia47 provides more limited evidence of effect. This study examined the provision of an advice session with an exercise specialist to sedentary adults (mean age 67 years). Although physical activity increased in both intervention and control groups, there was a significantly greater increase (p < 0.05) in self-reported frequency and time of vigorous exercise in the intervention group than in the control group (who received a nutritional pamphlet). Self-reported intention to exercise also increased significantly more in the intervention group (p < 0.0001). Although the paper reports these positive effects on vigorous exercise and intention to exercise, self-reported walking time and frequency of walking did not improve significantly more in the intervention group. However, a sample of the participants wore an accelerometer during the intervention period, and, with this measure (rather than self-reported data), there was no intervention effect on the daily number of steps. There was also no difference between groups with regard to physiological measures (such as body weight, blood pressure). Quality-of-life scores decreased for all participants over the 12-month study period, with the greatest decrease in women in the intervention group.

Costanzo and Walker34 provide weak evidence of effect following a behavioural counselling intervention. This study, which compared one session versus five sessions of counselling, reported an effect on self-efficacy, although no direct influence on physical activity. The change in self-efficacy (ES 0.19) was attributable to a decrease in the one-session group rather than change in the five-session group.

The other two papers reporting limited effects were the Marcus et al. 60 paper, which describes the pilot phase of work, and the Goldstein et al. 45 paper, which further outlines this. The evaluated intervention consisted of a physician training session and one physical activity counselling session provided to patients with one follow-up appointment 4 weeks later. Participants in the main study were an average age of 65 years and 36% were employed. At 6 weeks, participants in the intervention group were more likely to be in advanced stages of motivational readiness for physical activity than participants in the control groups [preparation or action 89% vs. 74%, odds ratio (OR) 3.56, 95% CI 1.79 to 7.08; p < 0.001]. However, the effect was not maintained at 8 months. There was no significant difference between intervention and control groups with regard to the number reporting meeting the recommended guidelines for physical activity at 6 weeks or 8 months. Both groups increased at 6 weeks and decreased by 8 months.

Group sessions

A total of 13 papers30,31,37,38,44,48,49,55,74,87–90 evaluated group-based programmes, with 11 of these reporting stronger evidence of effectiveness. 30,31,44,48,49,55,74,87–90 In total, four studies (reported in seven papers)30,31,55,87–90 compared differing forms of interventions rather than having no-intervention control groups.

The first of these, outlined in four papers,87–90 compared telephone-based counselling calls for 6 months to a 12- or 20-week group-based programme. Participants had a mean age of 68 years and were 80% female. The Wilcox et al. 87,88 papers present immediate follow-up data from one and three cycles, respectively, of the programmes. Results for the first implementation showed a significant increase from pre test to post test in self-reported moderate- and vigorous-intensity physical activity (p < 0.001) and total physical activity (p < 0.001) for all participants, with both programmes having significant effects. Across three cycles of the intervention, the telephone programme led to a significant increase in reported moderate and vigorous physical activity [Cohen’s d (d) = 0.62/0.66/0.75], and all reported physical activity also increased significantly (d = 0.55/0.6/0.63) in each of the 3 years in which the programme was run.

Similarly, for the group programme the study reported that moderate and vigorous physical activity increased significantly (d = 0.74/0.66/0.58), as did all physical activity (d = 0.79/0.56/0.63). For both programmes the proportion of participants reportedly reaching exercise recommendation levels increased significantly (p < 0.001). The papers present no data comparing effectiveness between the interventions beyond reporting the ESs for each, which were similar. The first Wilcox et al. 88 paper reports 6-month follow-up data and found a significant increase from pre test to post test for both programmes in self-reported physical activity, which was maintained at follow-up (p < 0.001). The second Wilcox et al. 89 paper analyses predictors of response to the intervention and reports that older participants (> 75 years of age) and those with less social support responded less well. It is noteworthy that those with lower activity levels at baseline improved the most.

Another study in this group that had two intervention arms30 assessed a programme which comprised access to a community exercise facility, a pedometer, peer mentors and group sessions. The participants were inactive and had an average age of 63 years, and 82% were female. The group sessions in one arm involved general health education, whereas the other arm received sessions on support for physical activity change. At the end of the intervention (16 weeks) both groups self-reported significantly more activity than at baseline [p < 0.001 (ES 1.38)], with no difference between the interventions. At 18-month follow-up, the intervention group receiving physical activity change sessions reported significantly more moderate to vigorous activity minutes per week than the general health education intervention arm [p = 0.04 (ES 0.32)].

Caperchione and Mummery31 similarly compared a standard intervention to an enhanced programme. The 12-week lifestyle intervention comprised instructor-led group walks and group education sessions incorporating cognitive–behavioural strategies and health-related information. The enhanced package also included education sessions on group process. Participants had a mean age of 58 years; 38% were retired and 50% were employed and were inactive. The study found a significant increase in self-reported physical activity behaviour (as measured by calorific expenditure) at time points up to 12 months’ follow-up for both groups (p < 0.05), with no significant difference between the standard and enhanced intervention.

The fourth paper comparing interventions55 assessed outcomes after interventions that included both group sessions and a home programme, in participants with a mean age of 68 years. One intervention focused on endurance and strengthening exercises and the other on stretching and flexibility. Although many of the outcomes of interest in this study are not specifically physical activity (e.g. bodily pain/endurance/strength), it included evaluations of walking distance via completion of exercise logs and self-efficacy via questionnaire, which are within the scope of this review. The endurance group reported greater improvement in the ability to walk a certain distance and self-efficacy (confidence) in walking (p < 0.03) than the stretching group.

An evaluation of a group intervention for retired women in the UK74 found that those who had taken part in the programme had significantly better self-perceived health, more exercise knowledge, positive exercise attitudes and more self-reported exercise than those who had not participated (i.e. those who had declined or who were not offered the course) at median 10 months’ follow-up. The group-based programme was taught for 2 hours a week by a trained Health Education Authority teacher and lasted for 10 weeks with the aim of promoting heart health. Following the intervention, the mean number of hours reportedly spent on sport/recreational exercise per week was greater in the intervention group than in the control group (p < 0.001). The time since completion of the programme did not have any effect on outcomes.

Another intervention aimed at older women was outlined in two papers. 48,49 Participants in this study had an average age of 65 years and were described as sedentary. The 26-week progressive walking programme was delivered in two sessions per week. At completion of the programme, participants were encouraged to continue physical activity. Following the intervention, the maximum current activity score had increased significantly from baseline (p = 0.01) in the intervention group compared with the no-intervention controls. At 6-month follow-up, 77.8% of the intervention group had reportedly continued with exercise and maintained lower exercise heart rates than the control group (p < 0.005). They also had maintained higher maximum current activity scores than the control group (p < 0.005).

A paper from Japan44 evaluated an exercise training programme for older adults (mean age 67 years). The 25-week programme consisted of three exercise classes each week with endurance training using bicycles and the use of flex-bands. Total daily energy expenditure (calculated from participant exercise logs) increased in the intervention group (from 40.8 to 43.5 kcal/kg/day, an increase of 6.4%; p = 0.03) and did not change in the control group (pre: 42.2, post: 39.2; p = 0.11). At 6 months’ follow-up there was some fading of effect (a decrease of 1.2 kcal/kg/day). Following intervention, reported daily energy expenditure remained higher than at baseline (p = 0.05). Reported moderate activity rates remained significantly higher, although strenuous activity had returned to pre-intervention levels.

Two papers from the same team37,38 provide limited evidence of effectiveness for a programme consisting of 30 once-a-week group sessions. The sessions comprised moderate-intensity recreational sport activities such as dance, self-defence or swimming led by a professional sport instructor. The mean age of participants was 59 years, and around of participants were female. Although the intervention resulted in significant increases on almost all questionnaire and fitness measures, similar effects were also found in the no-intervention control group. There was no statistical difference between the two study arms for sport activity, leisure time physical activity, health indicators or perceived and performance fitness measures. The intervention group improved significantly more than controls for only measures of sleep; diastolic blood pressure; perceived fitness; and grip strength. The second paper examines 12-month effects of the programme, although there is a substantial number of missing data (up to 49%). At this longer term follow-up, energy expenditure for sport and total energy expenditure had significantly increased from baseline (p < 0.01). However, the increase was not significantly different from the control group. Energy expenditure leisure time activity in the intervention group, despite increasing up to 6 months, decreased from 6 to 12 months’ follow-up. BMI and walking also significantly increased over time, but not when compared with controls. There was no significant difference by time or group for any of the other outcomes measured.

Individual exercise programmes

Seven papers (six studies)35,39,50,52,63,64,69 evaluated individual exercise programmes, with all providing evidence of effectiveness, although for one this was only at short-term follow-up. The first paper35 reported the results following supervised centre-based individual swimming or walking sessions, together with mini workshops prior to sessions for the enhanced intervention over 6 months. The programme continued unsupervised for a further 6 months. The study had four intervention arms (swimming and usual care, walking and usual care, walking + workshops and swimming + workshops), and participants had a mean age of 55 years. Benefit in terms of a reduction in time to walk 1.6 km was found for all groups at 6 months compared with baseline (p < 0.001), and this effect was maintained at 12 months (p < 0.05). The addition of the behavioural workshops did not increase the intervention effectiveness. Participants also improved in terms of distance covered during a 12-minute swim, with, perhaps unsurprisingly, the swim group having the greatest improvement in swimming time. The authors note that being older was associated with better adherence to the programme and retention (p < 0.05). Being in full-time employment was associated with adherence and retention at 6 months but was not associated with either at 12 months (p < 0.05).

Two other papers reported walking interventions. 50,69 Pereira et al. 69 examined long-term follow-up data following a walking programme for post-menopausal women (participant mean age 57 years). The original programme combined initial sessions of group walking for 8 weeks, with individual walking at least once a week to achieve 7 miles each week. After 8 weeks, participants could continue group walks or maintain solo walking. The intervention had been implemented in the early 1980s, with the authors impressively managing to collect 10-year follow-up data from 86% of participants. The median values for both self-reported usual walking for exercise (1000 vs. 302 kcal per week) and self-reported total walking (1344 vs. 924 kcal per week) were significantly higher for intervention participants than for controls participants (p = 0.01 for both outcomes).

A paper by Hekler et al. 50 drew a distinction between utilitarian walking (i.e. with the purpose of getting somewhere) versus leisure walking (i.e. for fitness or health). Participants in the study (mean age 64 years, 56% retired) received 2 weeks of each intervention, including instruction in the walking type, goal-setting, self-monitoring and problem-solving. Participants had a choice of walking type for a further 2 weeks. The study found a significant mean improvement in the amount of walking for all 2-week phases (p < 0.05). Significantly more mean steps per day (measured by a pedometer) were achieved by participants during the leisure intervention than during the utilitarian walking type (p < 0.05). The authors report that the order of instruction may have influenced this finding, as those instructed in leisure walking first had a greater difference in types of reported walking. All but two of the participants had preferred a mixture of walking types during the free-choice phase. The authors noted that neighbourhood characteristics were associated with the type of walking achieved, with more leisure walking undertaken if there was access to walking paths, better neighbourhood aesthetics, access to facilities and access to services. More utilitarian walking was reported by participants who travelled to multiple locations during a day.

Other papers in this group examined more general fitness rather than walking sessions. Hughes et al. 52 trialled a multiple-component physical activity programme which provided flexibility activities, low-impact aerobic exercise and resistance training for 60 minutes three times per week for 10 months. The study compared this intervention with any other intervention of the participants’ choice, including any other physical activity programme. Adults who participated in the multiple-component programme (mean age 66 years) showed statistically significant benefits at 5 and 10 months with regard to self-efficacy for exercise adherence over time (p < 0.001) and adherence in the face of barriers (p < 0.001). Self-reported frequency of all physical activity had increased by an average of 26% from baseline at 10 months in the intervention group compared with 9% in the control group (p = 0.028 baseline to 10 months). The difference in frequency of physical activity from baseline to 10 months, however, was not significant (p = 0.756).

A 14-week fitness programme evaluated by Dorgo et al. 39 led to improved fitness on all measures for both peer-mentored (p < 0.007; ES 0.2–1.6) and student-mentored (p < 0.31; ES 0.2–1.4) groups with no significant difference in outcome between groups. This study, which had no control arm, recruited volunteers with a participant mean age of 69 years. The authors noted that, although retention was high for both intervention arms, participation was higher in the student-mentored group.

There was one study63,64 that reported evidence of weaker effect following a fitness session intervention. This study, reported in two papers,63,64 had two intervention arms and a no-intervention control group, and participants were all retired with a mean age of 66 years. The second paper64 reports fitness, strength and body composition rather than physical activity outcomes. The interventions were either sessions at a fitness centre focusing on endurance, strength, flexibility and balance or an individualised home-based programme in which participants were encouraged to integrate physical activity into their daily routines. The home programme was supported by booster telephone calls from the instructors. The fitness sessions group significantly increased leisure-time physical activity and total physical activity from baseline to immediately post intervention compared with the control group, although the effect was not maintained at 23 months’ follow-up. The outcomes for the home intervention group showed significantly larger increases in active transportation, daily steps and total physical activity than in the control groups and had increased more than the fitness session group for active transportation and daily steps post intervention. At 23 months’ follow-up, the only measure that was significantly different between the control and intervention groups was self-reported kilocalories expended during active transportation in the home intervention group.

In-home telephone interventions

Nine papers (eight studies) evaluated in-home interventions which were predominantly delivered via the telephone. 32,51,56,57,61,62,72,73,78 The first of these papers51 had a no-comparator BA study design. Participants were a mean age of 68 years, and 49% were retired and not working, 17.7% were retired and working and 20.9% were employed. They received an initial meeting with a staff member to develop a physical activity plan and goals, followed by weekly then monthly telephone calls from a designated ‘buddy’. The study found a significant improvement in self-reported total calorie expenditure from baseline to mid-point and end point (p < 0.0001 median change 707 calories per week), with a typical increase of 3.2% of calories expended per month. Light and moderate self-reported activity levels showed a similar improvement (p < 0.0001 pre–post), and there was an improvement in the stage of readiness for all stages. Those with lower levels at baseline and who were older than 65 years of age tended to have greater increases in the numbers of calories expended.

Another similar intervention73 comprised an initial meeting to discuss fitness testing results and to develop a physical activity plan and goals, with 10 follow-up telephone calls over 6 months (it was unclear who made the telephone calls). The intervention group and control group both increased the weekly estimated calorific expenditure in all physical activity at 3 months’ and 6 months’ follow-up. The study provides weak evidence of effectiveness, as the intervention group had a larger increase, but this was not significant. There was no significant change in reported weekly frequency of activity or measures of physical fitness in either group. Moreover, there was no significant change in self-efficacy or perceived barriers. When controlled for baseline covariates (age, sex, income, BMI, support), intervention participants significantly increased the frequency of self-reported levels of physical activity compared with control participants (p < 0.05), suggesting that the intervention may have been effective for selected participants.

The third and fourth papers in this group61,62 examined up to 2-year outcomes following an intervention, which comprised a seven-session course delivered via the telephone by physical activity coaches. Participant mean age was 57 years, with 77% of participants employed. Significantly more activity was reported by the intervention group than the usual care control group at 6 and 24 months (p < 0.03 and p < 0.01). The intervention led to an increase in reported energy expenditure of around 200 kcal per week extra, equivalent to 1 hour per week of moderate-intensity walking. More intervention than control participants maintained moderate to vigorous physical activity at 6, 12 and 24 months (p < 0.003, p < 0.004 and p < 0.001). At 24 months, although the usual care group reverted to their 6-month level, the intervention group continued to increase (p < 0.05).

King et al. 56 compared human versus automated telephone contacts in participants aged over 55 years. The two forms of telephone intervention were compared with a health education class control condition. At 6 months, those in both the human advice and automated advice arms, although not significantly different from one another (p = 0.73), had significantly greater self-reported mean energy expenditure in moderate activity than the control arm (p = 0.01). Similarly, both the human advice and automated advice arms, although not significantly different from one another (p = 0.65), had significantly greater reported mean minutes per week spent in moderate activity than the control arm (p = 0.01). These differences were generally maintained at 12 months, with mean reported energy expenditure in moderate activity for both the human advice and automated advice arms not significantly different from one another (p = 0.60). These energy expenditure increases for both intervention groups were significantly greater than in the control arm (p = 0.036). Similarly, mean minutes per week spent in moderate activity for both the human advice and automated advice arms, although not significantly different from one another (p = 0.66), remained greater than in the control arm (p = 0.045).

Another study, which examined the potential of automated telephone systems,72 compared the effectiveness of a 12-week walking programme, including a pedometer and supported by an interactive voice response system, with support from a trained coach. The female participants had an average age of 57 years, and following the programme there was a significant improvement in 1-mile walk times pre–post (p = 0.001; d = 0.41). There was also a significant improvement in goal-setting, exercise planning and managing negative thoughts (ES 0.8). The ESs for BMI, weight and waist measurement were small (range 0.07–0.14). There was no between-group difference, with the addition of a human coach not seeming to increase effectiveness of the programme.

Three variants of a smartphone application (app) were trialled in an 8-week intervention. 57 Around half of the participants (48.5%) were in full-time employment, with a mean age of 59 years, and nearly three-quarters were female. The study found a significant increase in reported minutes of brisk walking for all groups pre–post (p < 0.0001), with no difference between the different apps (p > 0.73). The increase averaged 100.8 weekly reported minutes [standard deviation (SD) 167 minutes]. There was also an increase in reported moderate to vigorous physical activity for all groups (p < 0.0001), with no difference between them (p > 0.99). The app was described as being easy to use and motivating, with around half of participants willing to continue usage after the conclusion of the study.

Two studies compared telephone-delivered with mail-delivered interventions. 32,78 Castro et al. 32 examined whether or not these different methods varied in terms of encouraging maintenance. They randomised participants (mean age 56 years) to 1 year of telephone counselling which encouraged them to adopt either higher (more vigorous) or lower intensity physical activity. After 1 year, they were randomised to a second maintenance year of contact via telephone andmail or mostly mail only. The study found that for those individuals who had been in the higher intensity programme, exercise adherence was maintained better by mail. However, those who were in the lower intensity programme had better maintenance if they received both mail and telephone input. Motivational telephone calls were compared with print letters in a lifestyle intervention trial which targeted both healthy diet and physical activity outcomes, in participants with a mean age of 57 years. 78 The study found that the different intervention methods were equally effective compared with a no-intervention control group. ESs (Cohen’s d) for outcomes of interest including self-reported hours of physical activity per day ranged from 0.15 to 0.18.

In-home combined diet and exercise interventions

Six papers (four studies)29,42,43,65,83,84 outlined the results of interventions that targeted lifestyle more generally and consisted predominantly of materials delivered to the home by post/mail. A RCT of tailored activity and dietary newsletters was reported in two papers83,84 The participants were all female, with a mean age of 57 years. The paper evaluated two intervention arms, either generic or individually tailored newsletters, with no control group. Both groups improved significantly from baseline to 6 months on all measures of cardiorespiratory fitness and physical activity. At 12 months after completion of the intervention, the pre–post change for both groups was significant only for timed chair stands. At 12 months a significantly higher proportion of the tailored newsletter group than the generic newsletter group had reportedly achieved the activity behavioural outcome target of at least 210 minutes of moderate- or high-intensity activity weekly (30 minutes daily on 7 days each week) (p = 0.026). At 24 months (1 year following intervention completion), both groups had maintained reported levels of moderate or greater activity, stretching exercise and flexibility, but had declined in cardiorespiratory fitness. There was no significant effect across the three time points between the groups for any activity markers. Post hoc tests showed that the combined groups significantly increased the outcome of at least 210 minutes of weekly moderate or greater activity from 12 months to 18 months (p = 0.015). However, if considered separately, there was no significant change within either group.

Another study of a lifestyle intervention including both physical activity and nutrition compared a 6-month programme with no-intervention controls. 29 The sex of participants was balanced, with a mean age of 65 years, and 42% were retired. The multi-element programme comprised a booklet with recommendations which encouraged goal-setting, as well as an exercise chart, calendar, bimonthly newsletters, a resistance band and a pedometer. Telephone calls and e-mail contact were provided in accordance with participant request. In total, 74% of participants reported using the exercise chart, 62% practised the recommended exercises, 90% used the pedometer and 63% used the resistance band. Immediately post intervention, the study found significant improvements from baseline to post programme for the intervention group with regard to self-reported outcomes of strength exercise (p < 0.001), walking (p = 0.012), moderate activity (p = 0.008), vigorous activity (p = 0.044) and sitting time mean per week (p < 0.001). These improvements were significant compared with controls for strength (p < 0.001), walking (p = 0.0029), vigorous activity (p = 0.0015) and mean sitting time (p = 0.0001).

The third study considering combined diet and physical activity programmes provided participants with a personal health-risk report based on questionnaire data every 6 months, together with individualised recommendations, newsletters, self-management and health-promotion books and other materials, all of which were delivered by post. 42,43 The intervention was designed to cost around US$30 per year. The study compared the intervention arm to no intervention and completion of questionnaire-only arms. At 2-year follow-up, the participants who were all retired with a mean age of 68 years had reduced their health-risk scores by 23% compared with baseline in the intervention group compared with control (p < 0.001). As regards the two self-report physical activity measures (exercise minutes per week and exercise programme attendance), both groups had significantly improved baseline to 24 months (p < 0.001), despite the control group having declined in physical activity measures by the end of the first year.

One paper reported more limited effectiveness. 65 It evaluated the provision of a 6-month multi-element home-based intervention for insufficiently active adults (mean age 66 years), which included a booklet, guidelines, newsletters, a pedometer and a resistance band. It also included newsletters and support via telephone or e-mail contacts. The programme aimed to target both physical activity and nutrition outcomes. Although outcomes were more positive regarding improvements in diet, the effect on physical activity was limited. Immediately post programme, the reported time spent in moderate activity had significantly increased for the intervention group (p < 0.05). However, this had declined by follow-up, with reductions in mean walking time for both intervention and control participants.

Home-based interventions providing a pedometer/accelerometer

Although several of the studies reported in other groups included the provision of a pedometer as part of the intervention element, four studies evaluated home-based interventions in which the provision of a pedometer/accelerometer was the core component. 41,58,73,77 The first of these58 gave accelerometers to female participants (mean age 67 years). It was recommended that participants accumulated 9000 steps and 30 minutes of moderate-intensity physical activity per day during the 12-week intervention. Individual data regarding their steps over the previous 2 weeks were provided to participants, with recommendations made. A control group was also provided with an accelerometer to wear, which collected no data. At immediate follow-up, participants in the intervention group had increased their steps average by 16% (from 7811 to 9046 steps; p < 0.01). Moderate-intensity activity had also increased by 53% (95% CI 17.83% to 27.23%; p < 0.01). This was compared with no significant change in self-reported physical activity in the control group. Results on the walk time test also improved for the intervention group by 10% (p < 0.01).

A pilot study77 compared three intervention arms and a control group. The interventions were combinations of standard or individualised mailed educational materials and a pedometer. All variants of interventions led to an increase in daily steps for participants (mean age 64, mostly female) at the end of the 12-week intervention. The greatest increase was in the two groups which had the pedometer plus individualised booklets (significantly greater number of daily steps than groups 1 and 2; p < 0.001). The addition of biweekly telephone calls for one of these groups did not make any significant difference (p = 0.893).

Finkelstein et al. 41 explored the use of financial incentives combined with pedometer usage. Participants (mean age 59 years, 36% retired) were given a US$50 initial payment with subsequent payments dependent on the number of aerobic minutes achieved per day – nothing was given for fewer than 15 minutes, US$10 were given for 15–24 minutes, US$15 for 25–39 minutes and US$25 for 40 minutes or more. Daily use was capped at 75 minutes. The control group were given a payment for taking part and completing the study. The study found an adjusted treatment effect of just over 16 more aerobic minutes in the intervention group (p > 0.001). There was reportedly a small difference between full-time employees and retired people (with retired participants achieving more minutes), but this was not significant. The control group averaged 20 minutes of exercise per day, which showed a slight decrease over the 4-week intervention. The intervention group averaged 35 minutes per day (1.8 hours per week more than controls) with little change over the 4 weeks. Average pay-out to intervention participants was US$17.50 per week, which would equate to US$910 per year.

One paper73 trialled a 6-week activity programme, which was outlined to participants by a research assistant at a clinic visit. Examples of activities and general advice were provided, together with an activity monitoring log to complete. In addition, one group received a pedometer. This study was the only one in the included set that was carried out with a solely minority ethnic population. The participants were Native American elders with an average age of 58 years, of whom 21% were employed. At immediate follow-up, the reported frequency of walking had increased for both groups from baseline (p < 0.01), with improved calorific expenditure for all exercise-related activities (p < 0.001) and greater frequency of moderate-intensity exercise-related activities (p < 0.001) self-reported in both groups. There was no significant difference in effect between groups. The study therefore provides evidence of effectiveness for both interventions, with the effect of the pedometer not superior to advice only.

Computer-based interventions

Eleven papers (four studies)46,53,66–68,79–82,85,86 reported programmes using computers, all of which incorporate web-based components. Two of these studies report limited effectiveness. The first of these85 was the only paper we identified that referred to participants as recent retirees and recruited 413 mostly male participants with a mean age of 59.5 years. This Dutch cluster RCT evaluated an ‘energy balance’ intervention aimed to improve both physical activity and dietary behaviours. It consisted of five modules, with a pedometer, waist tape and information provided, delivered by post initially, then by CD-ROM, then by computer-tailored feedback. The fourth and fifth modules were available via the programme website, and newsletters were provided for the remainder of the length of the 2-year study period. The study authors reported limited effectiveness for the intervention, with similar improvements to physical activity and dietary behaviours in both intervention and control groups on most measures used. At follow-up, the change in reported sport and recreational activities was higher in the intervention group (74.5 minute/week) than the control group (23.4 minute/week; p = 0.03). Waist circumference, body weight and blood pressure decreased significantly in men in both the intervention and the control groups, but there was no significant between-group difference observed at 12- or at 24-month follow-up. The authors suggest that the study was underpowered to detect statistically significant differences and also highlight their recruitment of health-conscious individuals to intervention and control conditions.

A second paper46 described outcomes following the delivery of two newsletters (either tailored or standard) via the internet to women with a mean age of 56 years. A total of 6% were retired and 73% were in full-time employment. Although the majority of participants (83%) reported that they had read the newsletters, the study found no significant change for any self-reported activity or perceived benefits of activity at 3 months (1 month after receipt of the second newsletter). There was an improvement in flexibility pre–post (p = 0.02) and in levels of body fat. Both groups had improved in terms of reduced perceived barriers (p = 0.025).

The other two studies outlined more positive findings following internet-based programmes. Wijsman et al. 86 found that daily physical activity as measured by an ankle accelerometer increased by 46% in a mixed-sex intervention group with a mean age of 65 years (p < 0.001) compared with 12% in a wait-list control group (p < 0.001). There was also a small increase in moderate to vigorous activity in the intervention group (mean increase of 11.1 minutes per day) compared with a mean decrease in the control group (p = 0.001 relative difference) immediately following the 12-week monitoring and feedback programme. Irvine et al. 53 evaluated change at longer (6-month) follow-up subsequent to an internet programme which supported selection of activities and provided educational and supportive messages. The intervention group participants (mean age 60 years) showed significant improvement on 13 out of the 14 self-reported outcomes measures compared with a no-intervention control group immediately post intervention. Gains were maintained at 6 months, with all outcomes measures significantly different compared with the control participants (p < 0.001) including cardiovascular activities, stretching exercises, activity minutes per week, balance exercise, behavioural intentions to exercise and motivation to exercise.

Seven papers from one team in the USA explored and compared different aspects of similar programmes consisting of computer-delivered or print-delivered letters which either included (the plus intervention) or did not include (the basic intervention) environmental information in addition to personalised physical activity advice. The earlier papers from the team79,81,82 investigate efficacy and working mechanisms of the interventions. At 3 months, the printed letter intervention participants were 1.6 times more likely than wait-list controls to comply with guidelines (basic OR 1.67, plus OR 1.57), and at 6 months, intervention participants were 2.5 times more likely (basic OR 2.4, plus OR 2.8) than controls to comply with guidelines (p < 0.001). At 6 months, both intervention groups had enhanced awareness of their activity levels compared with controls (intervention OR 1.67, intervention plus environmental information OR 1.64; p < 0.001) with no significant difference between the intervention types. The intervention plus programme had a significant effect on cycling compared with the basic intervention (p < 0.01) and a significant effect on perceived physical activity possibilities (more than both controls and basic intervention; p < 0.05). This change was associated with a change in days of total physical activity (p < 0.05) and weekly minutes of sport (p < 0.001). The authors concluded that environmental perceptions were important as a mediating effect in changes in cycling, sport and total physical activity.

Efficacy of the basic intervention and the basic intervention plus environmental intervention was considered in a further paper. 80 This reported that both interventions significantly changed total self-reported weekly days of physical activity compared with the control group. Participants in the basic computer-tailored print intervention condition and intervention plus condition increased their total weekly days of sufficient physical activity from 4.3 days per week (SD 2.2 days per week) to 4.7 days per week (SD 2.0 days per week). Participants in the control condition had increased their total weekly days of physical activity from 4.0 days per week (SD 2.2 days per week) to 4.3 days per week (SD 2.0 days per week). However, only the environmentally computer-tailored print intervention significantly changed weekly minutes of physical activity. On average, participants in the intervention-plus-condition group increased their weekly physical activity behaviour by more than 1 hour per week (ES 0.19) compared with the control condition. Multiple mediation models indicated that the effects of both interventions on weekly days of physical activity were mediated by changes in awareness and intention.

In later papers from the team, the programme was developed into a web-based intervention. 66–68 Peels et al. 66,67 outline the development process and compare the new web programme to the original print-delivered information, in terms of participant types and attrition factors. The web version was designed to improve the reach and effect of the print version. However, the authors found that, although both high and low socioeconomic participants were reached by both versions of the intervention, delivery by print produced a higher response rate than the web version (19% vs. 12%). The web intervention appeared to have better participation rates among younger participants (the study included adults over 50 years, mean age 62 years), males and those with higher BMI. Dropout was also significantly higher in the web-based intervention (53% vs. 39%; p < 0.001). No specific participant characteristics were associated with dropout for the different interventions, with low intention to be physically active predicting dropout for both delivery modes (p < 0.001).

Effectiveness data compared a basic intervention via print or the web, plus environmental intervention via print or the web using self-report physical activity data. At 12 months’ follow-up the four intervention conditions as a whole were effective in increasing self-reported weekly days of sufficient physical activity (ES 0.18; p = 0.005) but ineffective in increasing self-reported weekly minutes of physical activity (ES 0.20; p = 0.071). The two paper-based types were equally effective in increasing total days and also total minutes of physical activity compared with control, whereas the two internet types were not. Neither web-based intervention significantly increased days or minutes of physical activity, although the control group decreased their physical activity over the same time period. Participant baseline intention to be active was the only factor that was associated with physical activity outcome.

Community-wide initiatives

We identified only two papers that evaluated physical activity interventions delivered across whole communities. 35,54 One reported little evidence of effectiveness. 54 This study54 from Japan recruited 12 communities with varying population densities. The intervention used social marketing principles and disseminated information via flyers, leaflets, newsletters, banners, used health professionals to provide encouragement during appointments and community events, and used community leaders and lay workers to provide encouragement and support via the provision of pedometers and reflective vests. The campaign was intended to target women aged 60–79 years with low back or knee pain, and the primary message of the campaign related to encouraging exercise in order to reduce knee and back pain. The study is of relevance to our review insofar as the authors predicted a ripple effect from the targeted segment to the wider community. The sample providing data were a mean age 60.5 years, around half were male and around 67% were employed. At baseline, 64.6% of the control group sample reported engaging in regular physical activity, compared with 63.9% of the intervention sample. At 1 year post campaign, the control group had decreased slightly to 60.3%, and the intervention group also decreased to 58.7%. The ES was not significant. There was some effect reported in terms of a significant difference in knowledge about physical activity benefit in control versus intervention participants at 1-year follow-up (OR 1.51; p < 0.05). The authors suggested that there had been contamination across study arms, as > 50% of control communities were aware of the campaign.

The second community initiative was the provision of free bus passes to over 60 year olds in England. The study35 analysed data from 3 years of the self-reported UK National Travel Survey to investigate any potential impact on having a bus pass and walking, cycling or travelling. This paper is of particular note in its examination of data relating to SES. Having a free pass was significantly associated with greater active travel among both disadvantaged (adjusted OR 4.06, 95% CI 3.35 to 4.86; p < 0.001) and advantaged groups (adjusted OR 4.72, 95% CI 3.99 to 5.59; p < 0.001), as well as with greater bus use in both disadvantaged and advantaged groups (adjusted OR 7.03, 95% CI 5.53 to 8.94; p < 0.001 and adjusted OR 7.11, 95% CI 5.65 to 8.94; p < 0.001, respectively) and a greater likelihood of walking more frequently in the whole cohort (adjusted OR 1.15, 95% CI 1.07 to 1.12; p < 0.001).

What are the most effective interventions to maintain and/or increase physical activity in older people during and shortly after the transition to retirement?

The majority of papers included in this review reported some intervention effect, although given that many studies reported multiple outcomes, including perceived change and readiness for change, this may be unsurprising. Six papers described change in related measures such as weight, fitness, self-efficacy or knowledge rather than direct physical activity. 34,45–47,55,60 It is important to note the tendency for improvement in participants over the period of the study which may not be related to the intervention. A significant proportion of studies used designs with several intervention arms, rather than a no-intervention control group and this obscures the level of intervention effect. There was evidence of the potential for a Hawthorne effect, with some studies finding improvements from baseline to follow-up in both intervention and control arms.

The Harvest plots below provide an overview of the effectiveness data by typology (Figure 7). In the chart each unique study is represented by a column, with the height of the column providing an indication of the strength of study design. The colour and pattern of each column provides further indication of study strength in terms of whether or not the study had a no-intervention comparator group and the strength of the outcome measurements used. As there was only a single study evaluating the education of health-care professionals, this has not been included in the plots.

FIGURE 7.

Harvest plots for A1/A2 intervention studies. Black, study with a control arm and measured data; dark green, study with multiple intervention arms only and measured data; light green, study with a control arm and self-report data; and blue, study with multiple intervention arms only and self-reported data. The height of the column indicates the strength of the study design. Studies in brackets have concerns regarding quality (see Appendix 2 for full details). (a) Advice/counselling; (b) individual session; (c) group sessions; (d) in-home computer; (e) in-home pedometer; (f) in-home telephone; (g) in-home diet and exercise; and (h) community.

The criteria for greater effectiveness versus lesser effectiveness were based on the proportion of outcome measures that were significantly different (p < 0.05 or p < 0.01), either from baseline to follow-up (for those with multiple intervention arms only), or between intervention and control groups. To be considered ‘more effective’, the majority of outcomes (at least half) relating to physical activity needed to show a positive intervention effect. Those studies categorised as ‘less effective’ reported few physical activity outcomes with positive effects.

The studies with the most rigorous designs are RCTs with control (no-intervention comparator groups) and including investigator-measured outcomes. These studies are represented in the graphs as the dark green columns. Studies in which there were more elements of concern in the quality appraisal assessment are in brackets.

As can be seen from the plots, the evidence regarding effectiveness for the different types of evidence is fairly positive across the set. This may be a result of publication bias. The approach to defining strength of evidence is a source of considerable debate within systematic reviews. A common method is to use the total number of papers of the highest quality RCT design as an indicator of strength, although this may merely indicate where more empirical work has been carried out. A refinement of this approach is to consider not only volume of papers but also consistency of findings.

In considering and comparing strength of evidence between the interventions, the pedometer-based studies all provided evidence of effectiveness, with three of the studies including measured outcomes (steps)41,58,77 and two using control group designs. 41,58 This suggests the strength of evidence for pedometer interventions in terms of consistency, although there are only four studies. 41,58,75,77 It is also noteworthy that these four studies all had only short or immediate follow-up and therefore do not provide evidence of a longer term impact for these interventions. One of these studies also indicated that, although the provision of pedometers could be effective, that pedometers did not produce superior results to advice only. 75

All but one38 of the group interventions studies seemed to result in positive effects, thus indicating strength in terms of consistency. 30,31,37,39,40,44,48,49,55,74,87–90 Half of the studies had control-group designs37,38,44,48,74 and also two used objective measures37,48 indicating strength in terms of quality. The evidence for this typology comprised a total of seven studies and therefore suggests strength of evidence not only in terms of design but also in terms of number of studies for group interventions. One of the more effective studies30 had a follow-up of more than 12 months, and all of the group had a follow-up period of at least 6 months, also suggesting the effectiveness of these interventions in terms of longer term change.

Most of the advice/counselling interventions also led to positive effects, although here there is less consistency, and only one study had a control group and measured outcomes. 47 Two of the three studies (three papers) suggesting less effectiveness also had greater risk of bias concerns. 34,45,60 In terms of follow-up, one paper70 had a follow-up of more than 12 months, and only one34 had a brief follow-up period.

All but one of the six studies evaluating individual session interventions were effective,35,39,50,52,69 although only one effective study had a control arm. 70 The single study (reported in two papers) suggesting less effectiveness was of a lower quality study design, although had more than 12 months’ follow-up. 63,64 One of the studies suggesting greater effectiveness69 also had more than 12 months’ follow-up. All but one of the in-home interventions provided predominantly via the telephone also appeared to be effective, with three of these studies having longer follow-up periods. 32,62,78 The study indicating less effectiveness73 was of the highest quality, although it had only a brief follow-up period.

Small numbers of papers make it difficult to draw firm conclusions regarding the effectiveness of the other types of interventions (in-home computer interventions, programmes combining diet and exercise and community programmes). The evidence here was also inconsistent, with equal numbers of papers suggesting both greater and lesser effectiveness.

Although the Harvest plot method of presenting the data provides a useful visual summary of effectiveness, it is important to note the detail contained in the previous narrative summary. The interpretation of the evidence is complicated by several studies comparing different interventions and often similar variants of intervention elements. The typology that we have adopted is one way of grouping the studies, and we recognise that alternative groupings are possible, particularly for those interventions with multiple elements.

A wide range of outcomes were considered to provide evidence of effectiveness within the set of papers. These included measured physical activity outcomes, measured biophysical outcomes and self-reported activity and biophysical outcomes. Few studies included assessment of sedentary behaviour. The two29,55 that measured inactivity highlighted the effectiveness of the interventions on reducing time spent sitting watching television.

In terms of applicability to the retirement transition period, we have previously highlighted the dearth of evidence specifically referring to this significant period of life change. All the studies, with one exception,85 set wide age ranges for inclusion or included only populations of retired people. However, there is no indication in the data that these interventions within the proxy age range that we adopted for the retirement transition (55–69 years) would be unsuitable or not effective for our target population. In the body of work that we included, we were also unable to find any evidence that the transition to retirement period was, or was not, a significant point at which to provide these interventions.

What is the evidence regarding the impact of interventions in different populations and/or the potential for retirement to increase health inequalities?

Differences in outcome between population subgroups were rarely mentioned by authors. One51 described more calories as being expended by older participants, another89 described older participants as achieving less improvement in physical activity than younger participants. This was described as potentially being associated with normal ageing in older adults, as with increasing age adults may have more health issues and less mobility. Another paper80 reported that including environmentally tailored information in letters was effective in changing total physical activity in those aged 64 years and younger, whereas there was no significant effect on total physical activity in older adults. The authors report that apart from this association, no other demographic factors appeared to be linked to the effectiveness of the in-home tailored letter intervention.

There was little evidence in the intervention literature regarding differential impacts of interventions on advantaged versus disadvantaged populations. The paper referred to above which explored potential associations between effectiveness and participant characteristics80 found no association between level of education and outcomes. A second paper66 mentioned that the web-based intervention had reached both high- and low-SES groups, and the authors noted that participants with higher levels of education requested more in-depth information regarding the impact of physical activity on health.

We have further examined issues of applicability of the interventions to our target population in the meta-synthesis of the qualitative and quantitative evidence outlined in Chapter 5.

The A3 studies

As described above, we developed an applicability rating for the identified literature, with those with a proxy age range most applicable to the retirement transition rated as A1/A2. We have outlined these in detail above (see What are the most effective interventions to maintain and/or increase physical activity in older people during and shortly after the transition to retirement?). We identified 39 additional papers with participants with a mean age 70 years or older, which were graded as being of lower applicability (A3) to our target population. Owing to the potentially more limited applicability, we will not be providing a detailed narrative synthesis of these papers. However, we provide characteristics of this literature in Tables 8 and 9 below, and we have summarised the effectiveness of these interventions via Harvest plots. We have provided additional detail for any studies of note and consider any similarities or differences between this group of studies and those in the A1/A2 category.

A3i papers (participants aged 70–75 years)

A3ii (participants aged 50–55 years)

What are the most effective interventions to maintain and/or increase physical activity in older people?

The interventions delivered to these participants did not differ from those carried out in the A1/A2 studies. The lower age group (A3ii) interventions included a similar range to that found in the slightly older population and encompassed counselling/advice; exercise classes; group or individual walking programmes; individual jogging programmes; and website-based interventions and information. The older group (A3i) similarly contained the same mix of advice/counselling, individual programmes and group programmes to that found in the A1/A2 studies. The different age of these participants did not, therefore, seem to have led to different intervention types or different targeting of interventions by age group.

As can be seen from Figure 8, in this set of papers there is a larger proportion of group-based interventions (15),91,92,96,98–102,112–115,120,124,129 perhaps reflecting the potential benefits of social interaction for elderly adults who may be living alone. In contrast with the A1/A2 studies, there are seven papers91,98–100,115,120,123 suggesting that these group programmes are less effectivene, and, therefore, there is less consistency in strength of evidence for group interventions in these populations. Apart from this difference, these studies with older/younger mean age participants echo the findings of the A1/A2 studies in tending to report positive outcomes for all intervention types. We have combined the in-home interventions that did not include computer-delivered programmes owing to small numbers and multiple elements. Most of these programmes included predominantly telephone advice/support; the Hughes et al. 128 paper included one intervention arm that consisted of web-based materials. There is strength of evidence for effectiveness of these in-home interventions in terms of volume of higher quality studies and in terms of consistency (all but two of nine studies indicated effectiveness). 93,95,105–107,116–118 Both studies121,122 in the group evaluating computer-based interventions report effectiveness, although this evidence is from weaker design, no-control-arm studies, and with self-reported outcomes.

What is the evidence regarding the impact of interventions in different populations and/or the potential for retirement to increase health inequalities?

As with the A1/A2 studies, we found very little evidence regarding differential effects among participants. One of the A3 studies126 was carried out in a wholly lower SES participant group and it found evidence of effectiveness for the mental contrasting intervention. The authors of this work highlighted an association between perceived importance of being active and physical activity outcomes. Seven of the A3 papers reported six studies carried out in a minority ethnic population. 91,94,99,105,106,115,120 These papers examined a range of intervention types (group, individual and advice) and found mixed evidence regarding outcomes. Four papers reported more evidence of effectiveness94,105,106,115 and three found less evidence of effectiveness. 91,99,120

Inclusion and exclusion criteria

Search strategy

Searches for qualitative papers were carried out using the same strategies as described for the effectiveness review. Papers that used qualitative methods were coded as such in the citation database, and those retrieved in the initial searches were used to identify keywords for the following iteration. Citation searches and author searches were carried out using relevant papers from these searches. In addition, the reference lists of included papers were scrutinised for new citations. Methods for citation sifting and retrieval are described in the methodology chapter.

Data extraction and quality appraisal

Data extraction was carried out using a modified form to include the most relevant data (see Appendix 7 for the extraction table). For qualitative papers this included summaries of the rich text from findings within each paper. We also noted the main theoretical underpinning of each study where this was discussed in the literature. Quality appraisal was carried out using an adapted version of the CASP tool for the assessment of qualitative studies (see Chapter 2). The reported details and methods of each paper were assessed for low or high concordance with items included in the quality appraisal tool.

Data synthesis

A large body of work was retrieved for the qualitative review, providing a challenge in terms of synthesising the data. We initially organised the studies according to age range/retirement status, specific physical activities, whether or not the activity was part of an intervention or a lifestyle choice, a focus on sex, ethnic issues or SES impact. As many studies focused on more than one of these aspects, we have, where necessary, included studies in more than one section of the findings. We were keen to highlight the impact of age, health, SES, ethnicity and sex with regard to physical activity. In order to identify potential strategies that might reduce inequalities, findings from the included papers were coded on the basis of these factors where they were apparent and analysed separately to create an explanatory narrative around these interactions.

Quantity of evidence available

Searches for qualitative literature produced 24 papers from a combination of initial searching and a further iteration (see Chapter 2 for details of the two searches). Five relevant papers were identified from citation searches and a further 26 from reference lists of included studies (see Figures 1 and 2). Grey literature was searched by our information specialist to identify any qualitative evaluations carried out by credible sources that were not published in journals; however, none was identified.

We retrieved a total of 55 qualitative papers reporting 48 unique studies. Thirteen studies reported in 15 papers were carried out in the UK. 130–144 The USA provided the largest body of research, with 18 studies reported in 22 papers. 145–166 A total of 11 studies were carried out in Australia167–175 or New Zealand,176,177 four were carried out in Canada,178–181 one in South Africa (producing two papers)182,183 and one in Chile184 (Figure 9).

FIGURE 9.

Number of studies from each country.

Quality of the evidence available

The quality of the included papers was generally variable, with 41 papers132,133,135–143,145–148,150–152,157,159–161,164,166–173,175–184 out of the included 55 at least partially meeting most of the eight quality criteria. There was a lack of reporting of the reasons for non-participation throughout the studies. However, many of the samples were small and purposive so this kind of information was not likely to be applicable. Ethical considerations were rarely addressed in a comprehensive way, with some papers reporting the details of ethical approval (where necessary) but not the considerations of ethical practice and vice versa. The third area in which reporting was often vague, brief or missing was the process of data analysis. Study authors may have been limited by journal word-count constraints. Limited detail meant that it was often difficult to assess how reported methods related to the actual implementation of the studies (see Appendix 4 for a quality assessment table).

Type of evidence available

Age

The age range of the study samples was broad and spanned the range 50–70 years. Three papers recruited individuals over the age of 70 years who were recalling past experiences of physical activity (Figure 10). 149,176,178

FIGURE 10.

Number of studies by age range of participants in years (not mutually exclusive).

Study design

In terms of methods used within studies, 21 studies reported in 23 papers132,133,135–139,146–149,153,159–161,171,174,176–178,180,181,184 used interviews exclusively to collect data (Figure 11).

FIGURE 11.

Number of studies by study design.

A further 13 studies used focus groups exclusively. 130,131,142,156,158,162–164,167,168,170,175,179 A total of five studies reported in eight papers134,143–145,150,157,166,173 used a combination of interviews and focus groups. A further nine studies reported in 11 papers140,141,151,152,154,155,165,169,172,182,183 used either interviews or focus groups and another qualitative method. In eight of these studies the additional method was participant observation. 140,141,151,152,154,155,169,172,182,183 The remaining study used focus groups and photo-voice. 165

Just over half of the included studies identified a theoretical background which was used as an explanatory tool for analysing the data. One study was based on a sporting model130 and five studies reported in six papers focused on sex/gender, with three studies (four papers)147,148,161,177 underpinned by feminist theory and two173,178 by masculinities theory. A further three studies that focused on ethnic minority experiences and views drew on theories of cultural sensitivity. 156,159,166 The most commonly identified theories were psychological/sociological theories, which were identified in 10 studies (13 papers)133,136,137,140,147,148,151,152,158,164,169,171,181 Ageing and life-course theories underpinned a further nine studies (11 papers). 135,141,145,146,149,150,155,157,172,174,180 Although the remaining studies (22 papers)131,132,134,138,139,142–144,154,160,162,163,165,167,168,175,176,179,182–184 did not explicitly mention any particular theory, a number drew on literature from ageing and sex/gender research when discussing the findings.

Sex

There was a predominance of female populations, with 18 studies reported in 22 papers136,137,145,147,148,153,154,156–161,163,165,166,169,172,177,181–183 recruiting all or mainly female participants. Four further studies155,173,175,178 as well as one paper from another study150 reported findings from male participants. A total of 24 studies reported in 26 papers recruited a balance of female and male participants or a mixed sample,130–135,138,140–144,146,149,151,152,164,167,168,170,171,174,176,179,180,184 and two studies did not report this detail139,162 (Figure 12).

FIGURE 12.

Number of studies including male/female participants.

Socioeconomic status

Ten included studies reported in 11 papers140,147–149,154,155,163,168,169,177,181 recruited samples with higher SES (defined by higher than average educational levels or income). Only two studies aimed to assess the physical activity experiences of older people from lower SES backgrounds. 160,172 The remaining 36 studies (42 papers) included a mixed SES population,130,132,134,141,142,145,150,157–159,161,164–166,170,174–176,180,184 or did not report SES characteristics131,133,135–139,143,144,146,151–153,156,162,167,171,173,178,179,182,183 (Figure 13).

FIGURE 13.

Number of studies by SES (education/income).

Ethnicity

Eleven studies reported in 15 papers133,145,147–152,154,157,161,168,169,178,181 included an all or mostly white population and seven included studies focused on ethnic populations, for example African American,162–165 African American and Native American159 or Latina. 160,166 Seven studies reported in nine papers recruited a mix of ethnic groups. 130,134,143,144,156,158,170,182,183 In the remaining 23 studies, reported in 24 papers, the ethnicity of the sample was not reported131,132,135–142,146,153,155,167,171–177,179,180,184 (Figure 14).

FIGURE 14.

Number of studies by ethnicity.

Activity level

Three included studies focused on currently inactive older people. 142,156,165 A total of 20 studies reported in 23 papers included individuals involved in or training others in a particular sport, activity or intervention, and therefore participants could be described as currently active. 135–141,149,151–155,161,169,171–173,176,179,182–184 In the remaining 25 included studies (29 papers), the activity levels of participants were mixed or not reported130–134,143–148,150,157–160,162–164,166–168,170,174,175,177,178,180,181 (Figure 15).

FIGURE 15.

Number of studies by activity level.

Intervention types

Four included studies reported in five papers135–137,139,140 explored individually prescribed interventions such as exercise referral schemes (ERSs). Ten studies reported in 11 papers140,149,151–154,156,167–169,184 explored views about a range of community-based interventions such as shopping mall walking or resistance training. Another two studies reported population-wide interventions,170,179 which in these cases were mass media campaigns. A further six studies reported in seven papers141,155,171,172,176,182,183 focused on a particular form of physical activity that the older research participants were regularly involved with. These were not health promotion interventions but were typically a sport such as tennis or bowling, or a dance activity such as ballroom or line dancing. The remaining 26 studies reported in 30 papers130,132–134,142–150,157–166,173–175,177,178,180,181 did not focus on a particular sport, activity or specific intervention; rather, they sought the views of older people about physical activity in general or potential barriers to participating in an intervention (Figure 16).

FIGURE 16.

Number of studies by intervention type.

The experience of retirement in relation to physical activity

Eleven of the included qualitative studies (reported in 12 papers)130–134,145,147–150,178 focused on or referred to physical activity and retirement. Nine papers mentioned retirement in the discussion of findings,138,146,151,152,155,157,169,173,177 although the aim of the study did not include a focus on retirement.

Individually prescribed interventions

Three included UK studies reported in four papers135–138 explored the views of older participants enrolled in GP ERSs or ‘Exercise on Prescription’ schemes and a further study examined views of professionals delivering the scheme. 139 Evans and Sleap135 explored the experiences of participants taking part in UK-based aquatic activities, many of whom had been referred to the intervention by their GP. Hardcastle and Taylor136,137 interviewed participants of a 10-week ERS for older women that had been in place for more than 10 years. Stathi et al. 138 interviewed a mixed-sex sample of ERS participants, each at different stages in one of three leisure centre-based programmes. Moore et al. 139 interviewed health professionals delivering a 16-week intervention based on motivational interviewing and goal-setting.

Participants relating their experiences of aquatic physical activity in one study135 were aware of the benefits of physical activity, but their understanding of their bodies was mainly in relation to regular routine biomedical measurements carried out at their general practice. The participants were thus keenly aware of the fact that their bodies were ageing and of the effects that this was having on their appearance. Aquatic physical activity differs from many other forms of exercise in that it requires exposure of the body, which was an inhibiting factor for many of the participants, particularly women but also some men. The participants mentally compared their bodies with media images of younger, fitter individuals and were conscious of how they appeared in bathing attire and to their contemporaries.

‘Exercise on Prescription’ classes were targeted at people over the age of 40 years, with activities modified to reduce intensity and therefore physical risk to the participants. These factors enhanced the sense of being an outsider in respect of the general population, and, because of the modified programme, of being ‘at risk’ even in the absence of health problems. However, a sense of shared experience with the group was forged and new support networks established. For some, the sessions offered empowerment through increasing achievements; the sense of ownership led to participants using strategies to continue attending when the prescribed 12 sessions were completed.

A majority of the women in another study reported in two papers136,137 initiated the idea of participation in ERS with their GP, having been told about the scheme by friends and peers. Some GPs were reported to have shown little enthusiasm about referral when the topic was raised, stating only that it might do them good. This use of initiative by the women signifies a pre-existing desire for change prior to discussing exercise with the GP. In contrast, men and women interviewed by Stathi et al. 138 depended upon their GP’s advice to encourage them to participate in and to continue with the programme. All but one health professional participating in one study139 reported that participation as a result of the patient’s own determination, rather than their GP’s advice, was more predictive of continuation with the programme. One health professional held the view that patients were more likely to attend the programme on their GP’s advice, as it was regarded as authoritative, compared with the advice of family. There was a suggestion that GPs target patients who were already motivated and that this might be a way to avoid wasted effort and reduce high dropout rates. Health professionals also reported that patients with mental health problems are difficult to recruit, as depressed patients have low motivation and some patients feel nervous about being watched. Similarly, people with a lower SES were perceived by staff as not engaging with the service. This was viewed by staff as perhaps being attributable to a lower value placed on health in an economically depressed community, in which unemployment rates are high, or to the fact that GPs are not motivated to put patients forward in these areas.

Women in one study reported in two papers136,137 reported that they faced stigma in terms of exercising in older age. There were reported perceptions from others that the women were trying to act or appear younger when they merely wished to improve their health and well-being. The authors refer to perceptions generally that older people may injure themselves if they exercise, or that they cannot live up to the ideal of being an ‘exerciser’. This potential anxiety was also raised in interviews with health professionals139 who suggested that patients might be put off ERSs by the presence of ‘fitter’ people at the gym. One suggested for overcoming this was to make it clear that the programme would be held at a time when other gym members were absent, as well as being clear about what is expected of patients when they attend.

Women in one study outlined in two papers136,137 reported some initial ambiguity towards participation, with some perceiving the appointment in a similar way to going to the dentist – they attended only because they felt they needed to and because they wanted to keep to their appointment. There were reports of continuing the programme out of a sense of responsibility towards their GP and also towards their own self-care to improve health. They had typically allowed physical activity to decline while looking after a house and children. At this point in their life they would have felt guilt at taking time for themselves. The women were seeking to re-enter a new phase in their life following retirement, which meant not only being more physically active but also more mentally stimulated, as well as extending social networks. For some women this new take on life increased their sense of commitment to the scheme so that other priorities were not given precedence: ‘Things are more important than doing housework and things like that’. 137 One woman found it difficult to fit the programme into her week initially, because her priority was looking after her husband. However, later in the programme she reported that it was no longer difficult; she was focusing on her own health and well-being, and exercise had become a way of life. Another woman had low motivation to begin with, but later in the programme she reported that the positive impact that her swimming had had on her well-being influenced her decision to continue.

The gym was reported to be a new environment for many ERS participants in all three studies, with feelings of discomfort expressed during the initial stages of a programme. It was therefore important for participants to receive encouragement and support from the instructor, particularly at the beginning of the programme. Health professionals were aware that vulnerable older adults might require extended support to promote feelings of security and to promote attendance. The implications of training for health professionals to deal with patients’ emotional issues and of the potential for patients to become dependent on this service were discussed. 146 For women in one study, the gym evoked an association with ‘exercisers’ with whom they did not at first identify. It would be easy to have been demotivated at this stage; one woman noted that her instructor’s ability: ‘. . . got me over the first hurdle’. 136

Although participants valued a secure environment in which to exercise, in one study138 some were not keen on structured exercise, preferring a combination of lifestyle or home-based physical activity that was meaningful as well as fulfilling exercise guidance. One man stated that he found the exercise schedule boring and would prefer to play golf. This difference in views between the two studies may be gendered, because one study interviewed only women136,137 who appear (from data in other included studies) to prefer participating in group activities more than men, perhaps owing to the potential to expand social networks. Health professional participants in one study139 also suggested that men may feel outnumbered by women, who tend to be referred more often, and that men benefit less from the social network. This would suggest that men have a different experience of ERSs to women.

Female participants in two studies136–138 were motivated to participate in ERSs for the social aspect. The participants in one study achieved this aim but Stathi et al. 138 reported that women did not meet new contacts at the gym, as they were attending only for a short period and participants often dropped out. For Hardcastle and Taylor136,137 social support and having somewhere to go were motivating factors, countering the sense of loneliness and not belonging that some had felt previously.

Women who committed to the scheme136,137 were reported to have adopted an ‘exercise identity’ which required habitual scheduling of exercise so that it became routine, reprioritising other tasks and developing a sense of achievement. This identity (which was reported to begin to develop from about 6 weeks into the programme) brought a sense of autonomy and control to women’s lives. The ‘exercise identity’ differs from exercise that is motivated by guilt and is therefore not autonomous. Women spoke of eventually feeling in control of their lives through exercising. Reported benefits of participation included feeling more energised, so that tasks at home were performed with more vigour. 136–138

To summarise, the evidence from included studies suggests that motivation to participate in ERSs could be enhanced through supportive recommendation from a health-care professional. However, it appears that an important motivator in maintaining exercise is an internal commitment to health and/or physical activity that may or may not be evident prior to referral. Motivation to participate can also be experienced differently across gender and socioeconomic groups. For example, women enjoyed the social element of physical activity where it was encouraged, whereas men tended to prefer less structured ways of exercising and could feel outnumbered owing to the low ratio of men to women referred. For lower SES groups and for women, at first there were concerns about beginning an activity that was associated with ‘exercisers’ and ‘fit’ people. Facilitators to attendance would, therefore, ensure an environment that includes only the ERS participants, with trainers that can identify with the individual, as well as with the gendered and social needs of participants. Trainers with a positive attitude who could motivate participants, particularly in early weeks, were particularly valued.

Community interventions to increase physical activity in people of retirement age and older

This section presents a synthesis of qualitative findings from ten included studies reported in 11 papers that examined views about interventions to promote physical activity at community level. 140,149,151–154,156,167–169,184 These included walking interventions, weight or resistance training, and fitness/dance/aerobic classes.

Mass media campaigns

Two included studies reported participant views about population-wide health promotion campaigns that included the aim of increasing physical activity in older people. 170,179

The studies170,179 found that recall of a Canadian and an Australian mass media campaign, respectively, was very poor. An advertisement that formed part of the Canadian campaign featured the grim reaper as a warning of what might happen if a healthy lifestyle was not taken up. Focus-group participants reported that this was positive as it attracted attention to the message, but negative in aspects of its content and its reminder to older people that they are people closer to dying rather than people who should be making the most of life. Participants held mixed views about whether or not the advertisement appeared to be aimed at older people. There was no verbal aspect to the advertisement, raising the comment that many people cannot read the message and would need to hear it. The advertisements were promoted by the government, a fact that was noticed by some participants. This led to reports of mistrust regarding messages that were deemed political. However, government sources were trusted more than commercial companies. Medical personnel were the most trusted source of information. Images of well-known retired athletes were mentioned as having the ability to engage people, although it was also mentioned that athletes are ‘not ordinary people’ and that regular members of the public would provide more realistic role models. Use of websites as further sources of information were criticised because not everyone has a computer or access to the internet. 179 Participants reporting on the Australian campaign suggested that health promotion messages should not be broadcast on television during the advertisement break as people tend not to view at those times. The recommendations to walk on most days were not perceived by the participants as being strong enough, with most people already carrying out that amount of activity. Television was not perceived as the most fruitful means by which to reach older people, with GP surgeries and newsletters preferred by this group. 170

In summary, recall of mass media campaign messages can be poor,170,179 although messages that are striking may be better remembered providing they are appropriate to the target audience. Campaigns can have a negative effect if messages are delivered in a way that is perceived as scare-mongering or paternalistic, or no effect if the audience is not able to access follow-up information. 179

The following section explores the views and experiences of retirement-age study participants of their involvement with physical activities that did not form part of an intervention.

Engaging with physical activity at retirement age

Six included studies reported in seven papers141,155,171,172,176,182,183 focused on older participants and their engagement with a range of physical activities. Three papers explored experiences and views about physical activity in general155,171,176 and three studies reported in four papers focused on a particular activity including lawn bowling,172 ballroom and sequence dancing141 and line dancing. 182,183

Factors relating to equality of access

A number of factors that impacted on access to physical activity in retirement age were identified from the literature. These are discussed by theme and summarised along with components of interventions that might limit such impact in Figure 17.

FIGURE 17.

Factors impacting access to physical activity in retirement.

Individual factors

Intervention factors

Barriers and facilitators to carrying out physical activity at retirement age

The preceding sections have presented factors that impact on decisions to engage with and continue physical activity at retirement age within the context of different activities and conditions. Table 20 presents a summary of the barriers and facilitators that have been covered by the sections, organised according to individual, intervention and interpersonal factors. Many of the factors cross cut and many are conditional on external influences such as interactions with partners and family.

The range of factors that might impact on participation in physical activity at this age suggests that a similarly broad range of interventions that take such factors into account may be required to encourage and maintain healthy activity levels. From the available literature, a particular challenge is encouraging initial participation for individuals who do not see themselves as ‘active’. This identity is further compounded by the effects of ageing and fears of injury. Many participants began participating through the encouragement of other people, such as family, friends, partners or health professionals, or out of a need to socialise or obtain some personal space. These factors may be at least as important as the health promotion messages that aim to increase participation.

Appropriateness of programme types

Programmes delivered in the home were most frequently identified by session participants as less relevant to people about to retire or recently retired. The types of programme that were perceived to be most preferred were group classes or individual training sessions (Figure 19). There was discussion around the importance of interacting with other people after retirement, particularly for those living on their own.

FIGURE 19.

Number of participants who identified a programme as preferred.

Factors influencing the amount of physical activity

Participants were asked to provide their opinion on whether or not each of the following factors (which were described in the qualitative literature) were important influences on the amount of physical activity that people participate in around the time of retirement (Table 22). As some respondents did not answer all the items, the numbers indicating positive and negative responses are provided for comparison.

Participants were also asked to identify the five most important factors influencing the amount of physical activity that people do around retirement, and the five least important influencing factors (Figures 20 and 21).

FIGURE 20.

Number of participants identifying a factor as an important influence.

FIGURE 21.

Number of participants identifying a factor as an unimportant influence.

There was considerable individual variation in views regarding the importance of the various factors. The elements more frequently identified as important related to keeping the mind and body healthy/active and meeting people. A feeling of guilt and lack of time were reported to be the least important influences on the amount of physical activity that people do around retirement age.

Factors influencing whether or not someone around retirement age would take part in a physical activity programme

The final area for input related to factors that might influence someone’s decision to take part in a physical activity programme. As above, some respondents answered only some of the items, so the positive and negative responses are provided for comparison (Table 23).

Most important and least important factors

As regards the most important perceived factors, the majority of items were selected by one to three participants, indicating a spread of views. Only the following elements were not selected by anyone: appropriateness to sex/gender or culture; stigma related to age; and GP has discussed the benefits. The most frequently selected items were: a wish to go out and meet people (n = 22); sessions providing a social element (n = 20); and free or low cost (n = 18).

There was little consensus regarding the least important elements, with many participants not selecting all five. Most items across the group were selected by at least one to three participants. The elements that were not selected by anyone (and therefore not considered least important) were: flexibly scheduled activities; embarrassment; enjoyment/fun; a social element; outdoors; gradual introduction; experienced benefits; instructor provides encouragement; and encouragement from others.

These sessions were carried out with only a small sample of participants, all of whom were already retired and were being asked to recall their views and experiences from an earlier time. The sessions also had the aim of further exploring the findings of the review rather than carrying out a primary research study. Similarly to the interventions studies, the sex ratio of respondents was uneven, with more women attending the clubs and groups than men. Given these limitations, the data provide further support for the findings of the review. The individual variation in rating of factors among respondents echoes the review in concluding that a range of interventions may be effective. It highlights that there is no one type of intervention that would suit all. The importance of a social element in interventions was apparent throughout these data, echoing the findings from our review of the qualitative literature. Other elements that were described in our review of the qualitative literature and highlighted in these data as being important in the design of any interventions included: low-cost interventions; interventions perceived as fun and enjoyable; flexibly scheduled interventions, although the sessions should have structure; activities underpinned by health advice from a trusted source; encouragement by family and friends; and personal invitations to take part could be beneficial.