Effectiveness of general practice-based physical activity promotion for older adults: systematic review

Abstract

Aim

To review the effectiveness of physical activity interventions for adults aged 50 and above, delivered through general practice.

Background

Physical activity has beneficial effects on the common disorders of later life. General practice is a potentially important setting for promotion of physical activity among older adults, but the effectiveness of such interventions is presently unknown.

Methods

Studies published between January 1998 and July 2011 were identified from electronic databases. We searched for studies of tailored physical activity interventions to older adults through general practice. The search and selection process was not restricted to any outcome measures but only included studies comparing two or more groups prospectively. Two reviewers screened the studies and obtained full texts of eligible studies. Included studies were assessed for their methodological quality and public health impact.

Findings

Altogether, 4170 studies met the initial search criteria but only six were included in the review, with a total of 1522 participants. The interventions ranged from six weeks to six months. One study showed a statistically significant increase in physical activity in the intervention compared with the control group (P ⩽ 0.007). Four studies measured quality of life using the SF-36, of which three reported inconsistent results. This review shows some evidence of the effectiveness of physical activity promotion for older adults through general practice, but not enough to warrant widespread commissioning and implementation. Large-scale developmental projects with long follow-up (beyond two years), objective measures of physical activity and comprehensive documentation of resource use, should now be conducted.

Introduction

Regular physical activity improves health and well-being. It reduces the risk of type 2 diabetes, osteoporosis, cardiovascular disease and some cancers, and falls in older adults (Baumann, 2004; Department of Health, 2011). Current recommendations are that each week older adults should do at least 150 minutes of moderate aerobic activity and two sessions of strength and balance activities (Department of Health, 2011). However, the older population in the United Kingdom is largely inactive (Skelton et al., 1999; Health Survey for England, 2009). Given population ageing, we need to promote regular physical activity in order to reduce the impact of disease, restore and maintain function, increase quality of life and contain the use of health and social services (Department of Health, 2011). There are both patient-focused and public health reasons for systematically promoting physical activity among older adults.

General practice-based physical activity promotion has the potential to change physical activity habits by addressing barriers to physical activity such as limited money and poor health (Kerse et al., 2005; Lees et al., 2005; Hardy and Grogan, 2009). Promotion of physical activity in general practice currently includes physical activity recommendations, written material and exercise referral schemes often based in local leisure centres, and there is some evidence that these approaches improve self-reported physical activity levels (Orrow et al., 2012). Much less is known about general practice-based tailored programmes that go beyond generalised advice. Therefore, this review evaluates the effectiveness of such general practice-based tailored physical activity interventions in older adults, whereby participants’ baseline physical activity levels are assessed to provide individualised physical activity recommendations. This review aims to be able to inform the commissioning and provision of physical activity promotion.

Method

Search strategy

We searched for studies that evaluated physical activity interventions for older adults using the following terms: (exercise* promotion or physical activit* or (strength and balance)) and (general practice or GP or general pract*) and (age* or older). Searches were run for research published from January 1998 to July 2011 in CINAHL Plus, EMBASE, MEDLINE, PUBMED, OT Seeker and Web of Knowledge.

Full texts of eligible studies were found and their reference lists were hand-searched for additional studies. Review papers were hand-searched to find the original articles. The PRISMA diagram shows the process of literature search (Figure 1).

Figure 1 PRISMA diagram. This PRISMA diagram shows the literature search results and the numbers of articles that were included and excluded from the review.

Study selection process

The two authors (Z.S. and C.B.) screened for eligible studies, any uncertainties were discussed between them and disagreements resolved by author S.I. Criteria for inclusion and exclusion were as follows.

Inclusion

• • Tailored physical activity interventions including aerobic, strength and balance exercises that recruited participants (aged 50 and over) from and/or were provided in general practice. ‘Tailoring’ in this review means baseline assessment of current physical activity and functional limitations, and individualised recommendations to increase physical activity.

Exclusion

• • Studies with participants with specific conditions (eg, advanced dementia or Parkinson's disease, frequent fallers, people with severe aortic stenosis).

• • Studies with participants recruited from care homes or not living independently.

• • Studies recruiting single sex populations.

• • Publications not in English.

• • Studies not comparing two or more groups prospectively.

Studies that involved participants below 50 years old were included if data were reported in separate age bands. Selection was not based on outcome measures.

Data extraction

Key details from the studies were extracted and entered onto a standard Excel grid with predefined headings (Table 1; Goldstein et al., 1999; Halbert et al., 2000; Petrella et al., 2003; Harrison et al., 2005; Kerse et al., 2005; Kolt et al., 2007). Data were extracted by one reviewer and checked by a second. Discrepancies were discussed and resolved between the two reviewers (Z.S. and C.B.) where necessary, involving reviewer S.I.

Table 1 Key characteristics of included studies

CHD = coronary heart disease; BMI = body mass index; CVD = cardiovascular disease.

Quality assessment and public health relevance

The studies selected for inclusion were assessed using two different sets of criteria, one addressing methodology and the other with relevance to public health.

Methodological criteria: the studies were evaluated for quality of controlled trials to assess for internal and external bias (Jüni et al., 2001). Internal bias includes selection bias, performance and detection bias and attrition bias. External bias includes generalisability of participants, treatment and setting.

Public health criteria: the RE-AIM framework allows for an evaluation of the public health impact of health promotion studies using five dimensions (Glasgow et al., 1999): (1) Reach: proportion of the target population reached and the characteristics of participants compared with the target population. (2) Efficacy: how the intervention benefitted the participants. (3) Adoption: characteristics of the settings participating in the study. (4) Implementation: the extent to which the intervention was delivered as intended, including the adherence to the intervention and the involvement of staff in the setting. (5) Maintenance: long-term maintenance of behaviour change, defined as equal to or more than two years.

Analysis

Meta-analyses were not performed because of the heterogeneity of outcome measures used by the studies.

Results

The literature search found 4170 studies. After review and exclusion of ineligible studies, six of these studies were included in this review (Figure 1). The six studies are described in detail below and summarised in Tables 1–3. Table 1 shows the key characteristics of included studies, Table 2 shows their methodological quality and Table 3 describes their public health impact according to the RE-AIM framework.

Table 2 Methodological quality of included studies

CVD = cardiovascular disease; CHD = coronary heart disease.

Table 3 Public health impact of included studies using the RE-AIM criteria

Description of studies

The six studies were all randomised controlled trials. Half of the studies were conducted in Australasia, two in New Zealand and one in Australia. The remaining three were based in the United Kingdom, United States and Canada. Five of the studies reported the mean age of participants, which ranged from 65 to 74; four recruited a greater number of females. Numbers of participants in the studies ranged from 168 to 355 and totalled 1522.

Four interventions were delivered through the general practice site (Goldstein et al., 1999; Halbert et al., 2000; Petrella et al., 2003; Kerse et al., 2005), one through a local leisure centre and one entirely by telephone (Goldstein et al., 1999; Kolt et al., 2007). Interventions were delivered by general practitioners (Goldstein et al., 1999; Petrella et al., 2003; Kerse et al., 2005), exercise specialists (Halbert et al., 2000; Harrison et al., 2005) and an exercise counsellor (Kolt et al., 2007). The interventions ranged from six weeks to six months. The number of contacts the participants had with the intervention deliverer varied between fortnightly (most frequent) to once in every two months (least frequent) (Halbert et al., 2000; Petrella et al., 2003; Kolt et al., 2007). The frequency of recommended physical activity varied because of the advice being tailored to individual participants: two studies encouraged participants to be active on two to three days per week (Goldstein et al., 1999; Halbert et al., 2000). The studies used a range of different outcome measures. Two studies used specific measures such as the PASE and the Auckland Heart Exercise Questionnaire (Harrison et al., 2005; Kolt et al., 2007). Three studies used self-reported activity and one study used a method of testing aerobic fitness using expired gas after exercising (Petrella et al., 2003). Secondary outcomes included Quality of Life using the SF-36 (Goldstein et al., 1999; Halbert et al., 2000; Kerse et al., 2005; Kolt et al., 2007), motivational readiness (Goldstein et al., 1999), blood pressure and falls (Kerse et al., 2005), self-efficacy and cardiovascular risk factors (Halbert et al., 2000; Petrella et al., 2003). The follow-up periods ranged from six to 12 months; four studies had 12 months follow-up (Halbert et al., 2000; Petrella et al., 2003; Harrison et al., 2005; Kerse et al., 2005).

Outcomes

Effects on self-reported physical activity levels

Two studies report a statistically significant increase in physical activity levels; Kolt et al. (2007) report that moderate leisure physical activity increased by 86.8 minutes/week in the intervention participants compared with controls (P = 0.007). More intervention participants reached 2.5 hours/week of moderate/vigorous leisure physical activity at 12 months compared with controls (42% versus 23%, OR 2.9, 95% CI 1.33–6.32, P = 0.007). Halbert et al. (2000) report that physical activity increased in both groups (P < 0.05), but more intervention than control participants increased their intention to do physical activity (P < 0.001). The increase was greater for the intervention than the control group for all measures except the time spent walking (P < 0.05; no odds ratio reported). Two studies showed no significant increase in activity (Goldstein et al., 1999; Kerse et al., 2005).

Effects on self-efficacy and motivational readiness for physical activity

Motivational readiness for behaviour change can be measured by the Transtheoretical Model for Change. Goldstein et al. (1999) showed that, at six weeks, 15% more intervention participants were in Preparation for behaviour change or Action phase compared with controls (OR 3.56, 95% CI 1.79–7.08, P < 0.001), and 16% more intervention participants improved from Pre-contemplation/Contemplation about behaviour change at baseline to Preparation or Action compared with controls (OR 3.27, 95% CI 1.32–8.07, P = 0.01). At eight months, no difference between groups in Preparation or Action was seen. Another study found physical activity self-efficacy significantly increased in intervention participants compared with controls at 12 months (P < 0.001; Petrella et al., 2003). Significantly, more intervention participants completed ⩾80% of available physical activity opportunities than controls (P < 0.05; no odds ratio reported).

Effects on aerobic fitness

VO_2max is a measure of aerobic fitness. In one study, VO_2max significantly increased in the intervention group compared with controls at six (11% versus 4%, P < 0.001) and 12 months (17% versus 3%, P < 0.001) (Petrella et al., 2003).

Effects on quality of life

One study found no differences on quality-of-life measures between the groups at 12 months (no odds ratio reported) (Kolt et al., 2007). Alternative research found a statistically significant decrease in the quality-of-life scores in both groups for body pain (P < 0.001), physical functioning (P < 0.001) and vitality (P = 0.04; no odds ratio reported; Halbert et al., 2000). Another study found that the intervention group showed significant improvements in vitality (OR 4.43, 95% CI 0.31–8.54) and general health (OR 5.46, 95% CI 1.69–9.24) scales compared with control (Kerse et al., 2005).

Other outcomes

One study reported that consultation time with participants was significantly longer with the intervention compared with control participants (P < 0.02) (Petrella et al., 2003).

Methodological quality

The methodological quality of the studies is summarised in Table 2. In terms of internal validity, only two studies describe the method for generating their randomisation sequence; one used a computer-generated sequence (Harrison et al., 2005), the other stated that it had been generated by an independent statistician (Kerse et al., 2005). Only two studies report on concealment of allocation (Petrella et al., 2003; Harrison et al., 2005). In five studies, those who delivered the interventions also assessed outcomes (Goldstein et al., 1999; Halbert et al., 2000; Petrella et al., 2003; Kerse et al., 2005; Kolt et al., 2007), thus blinding was not possible. Four studies did an intention to treat analysis (Halbert et al., 2000; Petrella et al., 2003; Harrison et al., 2005; Kerse et al., 2005), four studies controlled for confounding variables (Petrella et al., 2003; Harrison et al., 2005; Kerse et al., 2005; Kolt et al., 2007) and three studies reported no differences between treatment groups at baseline (Goldstein et al., 1999; Halbert et al., 2000; Petrella et al., 2003).

In terms of external validity, all studies took place in English-speaking countries, and half required participants to have a good understanding of English in order to participate (Goldstein et al., 1999; Kerse et al., 2005; Kolt et al., 2007). All studies aimed to exclude those who were regularly active and recruit sedentary participants. Four studies excluded people with unstable, progressive or debilitating illnesses (Halbert et al., 2000; Petrella et al., 2003; Kerse et al., 2005; Kolt et al., 2007). Four interventions were delivered in general practice, mainly by general practitioners but also by nurses and exercise specialists (Goldstein et al., 1999; Halbert et al., 2000; Petrella et al., 2003; Kerse et al., 2005). The interventions lasted from three to 12 months and were individualised for participants. Those who delivered the interventions had training for their role to provide participants with standardised information. One study provided participants with a leisure pass to attend a local Exercise Referral Scheme leisure centre for 12 weeks (Harrison et al., 2005).

Public health impact

The studies were evaluated using the RE-AIM criteria (Table 3):

• • Reach: Four studies report how many people were screened for eligibility or invited to participate and the percentage that consented, ranging from 10% to 67% (Goldstein et al., 1999; Halbert et al., 2000; Harrison et al., 2005; Kolt et al., 2007). The variation between studies could be partly because of the studies’ inclusion/exclusion criteria, determining the proportion of people who are included in these studies. The remaining two studies only report the number of participants who consented, hence the reach, and therefore the characteristics of participants compared with the target population is unknown (Petrella et al., 2003; Kerse et al., 2005).

• • Efficacy: The most common primary and secondary outcomes were physical activity level (Halbert et al., 2000; Harrison et al., 2005; Kerse et al., 2005; Kolt et al., 2007) and quality of life, respectively (Goldstein et al., 1999; Halbert et al., 2000; Kerse et al., 2005; Kolt et al., 2007). The results are varied; however, some interventions may have benefitted the participants in terms of their physical activity levels and quality of life. See outcomes in the ‘Results’ section.

• • Adoption: Five studies report the settings that hosted the interventions, which may be useful for future studies when assessing how interventions are adopted by settings. The number of general practice practices engaged in the studies varied from two to 42 (Goldstein et al., 1999; Halbert et al., 2000; Petrella et al., 2003; Kerse et al., 2005; Kolt et al., 2007). Three studies report the number of general practitioners who participated, ranging from 12 to 117 (Goldstein et al., 1999; Petrella et al., 2003; Kerse et al., 2005).

• • Implementation: All studies report the percentage of participants who completed the studies, ranging from 57% to 89%. Studies with lower participant attrition rates were delivered as intended more so than studies with higher attrition rates. All studies report who delivered the intervention; four studies report that the deliverer had training, background experience or qualification making them suitable for the position (Goldstein et al., 1999; Halbert et al., 2000; Petrella et al., 2003; Kerse et al., 2005).

• • Maintenance: None of the studies could be assessed for long-term behaviour maintenance because of follow-up periods of less than two years.

Discussion

Summary of main findings

The six studies included in this review were heterogeneous in design and difficult to compare. The most common outcome measures were of physical activity levels and quality of life. One study, providing three months of physical activity fortnightly counselling over the telephone, found a statistically significant increase in physical activity in the intervention participants compared with controls (Kolt et al., 2007). A study found statistically significant increases in quality of life scores for vitality and general health in participants who received a ‘Green prescription’ (exercise on prescription) and telephone follow-up (Kerse et al., 2005). One study showed a statistically significant increase in aerobic capacity in the intervention participants compared with controls (Petrella et al., 2003). Four studies used general practitioners or nurses to deliver interventions (Goldstein et al., 1999; Halbert et al., 2000; Petrella et al., 2003; Kerse et al., 2005); however, methods of assessing outcomes and follow-up periods differ. Half of the studies gave participants practical encouragement for increasing physical activity, either by providing a membership to a health centre (Harrison et al., 2005), by providing a step monitor (Petrella et al., 2003) or by giving specific physical activity plan (Goldstein et al., 1999; Halbert et al., 2000; Petrella et al., 2003; Kerse et al., 2005). All interventions left participants to motivate and organise their own physical activity, and the quantity of physical activity undertaken was not monitored, making it difficult to know whether the dose of the intervention affected the results.

The methodologies of studies were not well reported, making it difficult to replicate the interventions and determine their quality. The risk of selection bias is high in the four studies that have no mention of allocation concealment (Goldstein et al., 1999; Halbert et al., 2000; Kerse et al., 2005; Kolt et al., 2007). The risk of detection and reporting bias is high in five studies where blinding was not possible (Goldstein et al., 1999; Halbert et al., 2000; Petrella et al., 2003; Kerse et al., 2005; Kolt et al., 2007). The risk of attrition bias is low in three studies that controlled confounders and performed intention to treat analyses (Petrella et al., 2003; Harrison et al., 2005; Kerse et al., 2005). The studies were undertaken in different countries, in rural and urban settings. Follow-up was less than two years, making it difficult to know which interventions have sustained effects.

Strengths and limitations of the review

This is the first systematic review to contextualise physical activity promotion for older adults using the RE-AIM framework, to provide a public health perspective.

This review has a narrow scope. Only studies set in general practice that recruited healthy participants were included, and all eligible studies only recruited English-speaking participants in developed countries. The review excluded studies that concentrated on physical activity as therapy for specific medical conditions or syndromes, for example, falls. Because of limited research in this area, no restrictions were put on outcome measures in selected studies. Inevitably, a main limitation is heterogeneity in the interventions and outcome measurements. All studies had self-reported physical activity outcomes, sometimes captured using a standard instrument; self-report may overestimate levels of physical activity (Hillsdon et al., 2005).

Participants who choose to take part in physical activity studies are likely to be more active or over-report their physical activity levels compared with the wider population (Goldstein et al., 1999). This potential selection bias may have occurred in these studies, but the inclusion criteria minimised this by excluding regularly active people. It is impossible to blind participants to their given physical activity intervention. Reporting bias, therefore, may have occurred in the reviewed studies, which would make the interventions appear more effective than they actually are. Despite small sample sizes, four studies reported their power calculation and recruited enough participants to detect differences (of varying levels) with 80% or 90% power at a 5% statistical significance (Table 1; Halbert et al., 2000; Petrella et al., 2003; Harrison et al., 2005; Kolt et al., 2007).

Implications for future research and clinical practice

The evidence for the effectiveness of general practice-based physical activity promotion aimed at older adults is too limited to support widespread commissioning of such interventions. It does suggest that large-scale developmental projects with follow-up periods exceeding two years, objective measures of physical activity and evaluation of service used to determine the implications for clinical practice should now be considered.