The clinical effectiveness and cost-effectiveness of exercise referral schemes: a systematic review and economic evaluation

Article history

The research reported in this issue of the journal was commissioned by the HTA programme as project number 08/72/01. The contractual start date was in September 2009. The draft report began editorial review in December 2010 and was accepted for publication in March 2011. As the funder, by devising a commissioning brief, the HTA programme specified the research question and study design. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the referees for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

JS is Chief Medical Officer of the Fitness Industry Association (FIA). The FIA meets his receipted expenses. The post attracts neither a salary nor fees.

Copyright statement

© Queen’s Printer and Controller of HMSO 2011. This work was produced by Pavey et al. under the terms of a commissioning contract issued by the Secretary of State for Health. This journal is a member of and subscribes to the principles of the Committee on Publication Ethics (COPE) (http://www.publicationethics.org/). This journal may be freely reproduced for the purposes of private research and study and may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NETSCC, Health Technology Assessment, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

2011 Queen’s Printer and Controller of HMSO

Physical activity and health

Physical activity (PA) contributes to the prevention and management of over 20 medical conditions and diseases, including coronary heart disease (CHD), stroke, type 2 diabetes mellitus, chronic back pain, osteoporosis, cancers, falls in the elderly, chronic obstructive pulmonary disease (COPD), decline in physical and cognitive function, depression and dementia, as summarised in the Chief Medical Officer’s (CMO) report At Least Five a Week: Evidence on the Impact of Physical Activity and its Relationship to Health1 and, more recently, the US guidelines for physical activity. 2 Table 1 lists some of the key conditions in which exercise has been shown to be beneficial. The CMO report estimated the total cost of physical inactivity in England to be £8.2B per year.

Current recommendations are for adults to aim to be active daily. Over a week, activity should add up to at least 150 minutes (2.5 hours) of moderate-intensity activity in bouts of 10 minutes or more – one way to approach this is to do 30 minutes on at least 5 days a week. Emerging evidence on the effects of time spent in sedentary activities (e.g. television viewing) on obesity, metabolic processes and type 2 diabetes, independent of PA, suggests that reducing time spent in sedentary activities may be an additional useful indicator of the effectiveness of interventions. Worldwide, over 20% of CHD4 has been attributed to physical inactivity, and the most active are at 30% lower risk for developing CHD than the least active,5 with a stepped reduction in risk. The dose for reducing risk in respect of other diseases and for promoting positive well-being is less clear, but the minimum target has been widely recommended as being reasonable for general health benefit at population level. 3

The Health Survey for England (HSE) provides national data on PA prevalence in England. The 2008 HSE report estimated that 39% of men and 29% of women meet the public health target of 5 × 30 minutes per week, with evident variations across age, sex, class and ethnicity. 6 The proportion achieving the targets for PA appears to have increased from 32% in 1997 to 39% in 2008 for men and from 21% to 29% for women. Nevertheless, there is a clear need to promote PA, particularly among the least active, who may have most to gain in terms of health. For adults, efforts in promoting PA have focused on changes in the environment (e.g. walking and cycle paths),7 mass media campaigns, web- and information technology-based communications at population and individual level,8 corporate and workplace initiatives,9 community programmes,10 and provision of individualised professional support11 and new health-care structures. 12

Reviews have also focused on the effectiveness of different PA interventions among specific groups in the population, such as the elderly13 and workers. 14 Systematic reviews suggest that no single approach can be wholly effective8 in helping sedentary people to maintain a physically active lifestyle, and that a wide variety of approaches can each facilitate small increments in behaviour change. The Foresight report on obesity15 reflected determinants of PA with regard to its influence on energy balance, and this is reflected in the cross-governmental policies in transport, health, schools and the built environment to tackle it.

Theories of behaviour change also support the need for multiple-level (e.g. targeting attitudes of both recipients and providers of health promotion messages) and multicomponent approaches (e.g. targeting different belief and attitudinal dimensions, such as the importance or salience of new behaviours, confidence to change, expectancy of benefits and beliefs of others). 16 The past 15 years have seen a growth in the understanding of physically active behaviour and in how to promote it with strategies matched to individual needs. 17 Achieving and maintaining a physically active lifestyle may require numerous and diverse changes in how individuals interact with the environment and with others, as well approaches such as self-monitoring of PA. 18 In evaluating the effectiveness of interventions, it is important to understand precisely what the intervention was and whether this was achieved, and also what process or mediating variables were implicated in changes in primary outcomes (i.e. behavioural and health outcomes). Many reviews and individual studies report the behavioural outcomes or biomedical markers yet very few describe the processes involved in behaviour change. 19

Physical activity promotion in primary care

Primary care has been recognised as a potentially important setting for the promotion of PA. 20 Over 85% of the population in the UK visit their general practitioner (GP) at least once a year, and almost 95% do so over a 3-year period,21 suggesting an opportunity to promote PA. Taylor22 identified, in a review of literature, several barriers that GPs perceived in promoting PA: (1) lack of time in the course of normal clinical interactions in primary care; (2) a lack of desire to pressure patients; (3) a belief that it may not be as beneficial as other therapies or other behavioural targets (e.g. smoking); (4) that patients would not follow advice; and (5) that PA promotion often seemed irrelevant for the needs of patients at the time of consultation. 22

Within the primary-care setting, there are broadly two models of PA promotion – exercise recommendation and EFSs. Although often referred to interchangeably, there are important differences between the two models:

1. Exercise recommendation Within the exercise recommendation framework, primary-care practitioners identify inactive adults and directly offer the advice or counselling regarding exercise, and/or a written prescription of exercise. In its guidance on PA promotion,23 the National Institute for Health and Clinical Excellence (NICE) recommended that a validated tool, such as the Department of Health GP Activity Questionnaire (GPPAQ24), be used to identify inactive adults. Boxes 1 and 2 summarise the intervention description from two randomised controlled trials (RCTs) that illustrate the model of exercise recommendation.

2. Exercise referral schemes As in the exercise prescription approach, inactive adults are identified in the primary-care setting. In this case, instead of direct PA advice, the GP or health-care professional refers the patient to a third-party service, with this service taking responsibility for prescribing and monitoring an exercise programme tailored to the individual needs of the patient. NICE defines an exercise referral scheme (ERS) as a process whereby a health professional ‘directs someone to a service offering an assessment of need, development of a tailored PA programme, monitoring of progress and a follow-up. They involve participation by a number of professionals and may require the individual to go to an exercise facility such as a leisure centre’.

Within this intervention model, the third-party service may often involve a referral to a local sport or leisure centre. However, the model can also include referral to a practice-based exercise specialist or physiotherapist. The interventions of two recent RCTs evaluating ERS are detailed in Boxes 3 and 4.

Some trials have been conducted that have evaluated a primary care-based PA promotion including elements of both exercise prescription and ERS. One particular example is The New Zealand ‘green prescription’ (Box 5). In this model, the GP prescribes for the patient an exercise programme, and advises the patient that telephone support is available from the local sports foundation, if required. The failure to differentiate between different models has led to ambiguity within the literature, with different interpretations of these models, particularly in systematic reviews (see Chapter 3, Quality of previous systematic reviews, Scope of previous systematic reviews, and Findings of previous systematic reviews, for further discussion).

Development of exercise referral schemes in the UK

Formal links between health care and promoting healthy living through opportunities to exercise are not new. For example, the Peckham Health Centre, in south London, was a bold departure in the medical field in the 1930s, concentrating on a preventative rather than a curative approach to health. To facilitate their grand project, two doctors housed in this purpose-built building engaged with over 900 families as part of ‘the Peckham Experiment’. For one shilling (£0.05 in today’s currency) a week, they relaxed in a club-like atmosphere: physical exercise, games, workshops or even simple relaxation were all encouraged.

The first contemporary ERS was set up around 1990, and over the past two decades there has been a significant and sustained growth in the number with possibly over 600 ERS operating across the UK. This rapid growth in the number of ERS has occurred, in part, in response to new legislation (e.g. compulsory competitive tendering and private management30) of such facilities. Leisure centres with swimming pools and other exercise facilities provide the opportunity to offer diverse options, as well as providing social facilities and became more business orientated, and broadening their clientele base and selling more direct debit-type memberships instead of ‘pay as you go’. The first evaluation of schemes was commissioned by the Health Education Authority in 1994. 31

In the 1990s, several limitations in ERS were indentified:32,33 (1) there were few of them, so they had little potential to impact on public health; (2) staff lacked the training to adapt exercise programmes to the specific health needs of patients; (3) there was little interest in the broader promotion of a more physically active lifestyle, but more interest in building leisure centre membership numbers; (4) GPs were reluctant to refer patients to exercise professionals who had unknown expertise and credentials; (5) there was only limited reference in key NHS policy documents to the promotion of PA; and (6) schemes were inadequately resourced for long-term evaluation. 34 As a result, and after broad consultation with health and exercise professionals, leisure industry operators and exercise scientists, a National Quality Assurance Framework (NQAF) was launched in the UK in 2001 to guide best practice and best value from ERS. 12 The document was aligned with the emerging range of NHS National Service Frameworks (e.g. for CHD, older people) that prioritise PA promotion.

The NQAF12 recommended a service-level agreement to drive the operational links between the primary-care referrer and the exercise or leisure provider, with exercise professionals on the Register for Exercise Professionals (www.exerciseregister.org/) at least at a level compatible with the needs of their clients (level 3: Instructing Physical Activity and Exercise). National Occupational Standards for level 4 (Specialist Exercise Instructor) in Health and Physical Activity were developed in 2007, with core units for CHD, mental health, obesity/diabetes, frailer older adults/falls prevention, after-stroke care and back pain. Despite the publication of the NQAF, capacity and resource constraints have largely dictated the extent to which schemes are meeting these standards. Furthermore, researchers have argued that the NQAF has failed to achieve consistency and comparability of standards, audit and evaluation mechanisms across the country. 33

The most recent survey of ERS programmes was undertaken by the British Heart Foundation National Centre for Physical Activity and Health (BHFNC34), from September 2006 to February 2008. In total, 158 schemes from England and Scotland provided information for the survey. Among these schemes, reported referral rates ranged from 20 to 6500 patients per year. Reported uptake rates (patients attending the initial consultation) ranged from 30% to 98%, with 82.5% of schemes having follow-up system for patients not attending initial consultations (telephone calls, letter/postcard). Scheme completion rates ranged between 20% and 90% depending on the ‘completion’ measure used. Although 95% of schemes reported collecting routine adherence data, adherence levels were not reported as part of the survey.

Fifty per cent of schemes had PA-based inclusion criteria, varying from less than 30 minutes’ activity per week or < 5 × 30 minutes of activity per week, with others using PA questionnaires (e.g. GPPAQ) to determine activity levels. Most schemes received patients with a range of medical conditions, including hypertension, weight problems, diabetes, arthritis, osteoporosis, anxiety and depression (see Appendix 1, Figure 21).

The survey reported further information regarding ERS in the UK that included how long schemes have been running; the aims of the scheme; the scheme characteristics (facilities and activities); the length of referral period (in 47% of schemes this was a 12-week period); and the extent to which the NQAF was used to inform the scheme. However, it was acknowledged that this information provided only a snapshot of operating EFSs, as an estimated 64% provided information.

Current guidance on exercise referral schemes in the UK

In 2006, the NICE Public Health Intervention programme undertook a review of the effectiveness of brief primary care-based intervention for PA promotion that included ERS. 35 NICE determined that there was insufficient evidence to recommend the use of ERS as an intervention to promote PA, other than as part of research studies where their effectiveness could be evaluated. NICE guidance35 recommended that:

Practitioners, policy makers and commissioners should only endorse exercise referral schemes to promote PA that are part of a properly designed and controlled research study to determine effectiveness. Measures should include intermediate outcomes, such as knowledge, attitudes and skills, as well as a measure of PA levels. Individuals should only be referred to schemes that are part of such a study.

Following NICE guidance, and in consultation with exercise referral professionals, commissioners and referring practitioners, the BHFNC published A Toolkit for the Design, Implementation & Evaluation of Exercise Referral Schemes. 34 As noted by its authors, this toolkit is not meant as a replacement for NICE or NQAF guidance, but aims to provide a set of guidance on the implementation and evaluation of ERS for referring health-care professionals, exercise referral professionals and ERS commissioners.

Summary

• Physical activity contributes to the prevention and management of a number of medical conditions and diseases.

• Currently, only 25–40% of adults in UK meet the CMO’s target for PA.

• Primary care is a potentially important setting for the promotion of PA, resulting in the ERS model being developed.

• Although variations in the model of delivery in ERS across the UK exist, common features include (1) identification of sedentary individuals at risk of lifestyle diseases by a health-care professional operating within a primary health-care setting; (2) referral to an exercise professional who seeks to develop a programme of exercise tailored to the needs of that individual patient; (3) monitoring of progress throughout the programme with appropriate feedback to the referring health-care professional; and (4) auditing to ensure adherence to quality assurance processes (e.g. appropriate staffing, health and safety procedures, ethical and data protection consideration).

• Despite a NQAF for ERS, capacity and resource constraints have largely dictated the extent to which the majority of schemes are meeting these standards.

• The NICE guidance in 2006 concluded that there was insufficient evidence to recommend the routine use of ERS to promote PA and called for further clinical effectiveness research.

Decision problem

• Interventions For the purposes of this report, an ERS was defined as comprising the following three core components:

  1. – referral by a primary-care health-care professional to a service designed to increase PA or exercise

  2. – PA/exercise programme tailored to individual needs

  3. – initial assessment and monitoring throughout the programme.

• Population including subgroups The population for this study was defined as people with a diagnosed condition known to benefit from PA. [Although the commissioned scope of this report was to focus on those with a diagnosed condition (known to benefit from PA), given the lack of evidence in this population, we broadened the scope of this report to include individuals without a medical diagnosis.] Subgroups of interest will be identified by diagnosed condition.

• Relevant comparators All relevant comparators were considered including usual care (e.g. PA advice or leaflets), an alternative form of PA intervention or different forms of ERS.

• Outcomes All relevant outcomes were sought. Given the nature of the intervention, we were particularly interested in changes in PA. PA can be assessed in number of ways [e.g. self-report or objective measures of PA, proportion of people meeting guideline recommendations, minutes per week of PA (total or moderate intensity), energy expenditure] and we considered all of these approaches. Other outcomes sought were uptake and adherence to ERS, physical fitness, clinical outcomes (e.g. blood pressure, serum lipids), psychological well-being, health-related quality of life (HRQoL), patient satisfaction, and potential adverse events of ERS (e.g. musculoskeletal injuries).

Overall aims and objectives of assessment

The overall aim of this review was to assess the clinical effectiveness and cost-effectiveness of ERS in people with a diagnosed condition known to benefit from PA. [Although the commissioned scope of this report was to focus on those with a diagnosed condition (known to benefit from PA), given the lack of evidence in this population, we broadened the scope of this report to include individuals without medical diagnosis.]

This aim is addressed through undertaking:

• a systematic review of the clinical effectiveness of ERS

• a systematic review of published economic evaluations of ERS

• a systematic review to identify predictors of ERS uptake and adherence

• the development of a decision-analytic model to extend published results and to generate expected values for the health and cost gains/losses associated with ERS.

The specific objectives of the review are to:

• assess the clinical effectiveness of ERS (see Chapter 3: includes individuals without a medical diagnosis – see note in parentheses above)

• assess the cost-effectiveness of ERS (see Chapters 4 and 6: includes individuals without a medical diagnosis – see note in parentheses above)2

• identify predictors of uptake and adherence to ERS (see Chapter 5)

• explore the factors that might influence the clinical effectiveness and cost-effectiveness of ERS (see Chapters 3 and 6: includes individuals without a medical diagnosis – see note in parentheses above)

• identify priorities for future research in this area (see Chapters 7 and 8).

Methods

This clinical effectiveness review was conducted and reported in accordance with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement. 36

Results

Identification and selection of studies

Our bibliographic search yielded 21,563 titles, of which seven primary studies and five systematic reviews were judged to meet the inclusion criteria. Figure 1 summarises the selection and exclusion process.

FIGURE 1.

Study inclusion process for ERS effectiveness systematic review. a, Primary exclusion criteria identified. PHC, primary health care.

Primary exercise referral scheme studies

As shown in Figure 1, the most frequent reason for exclusion from the present review was that studies used primary care as means of recruiting individuals into exercise programmes, but there was no clear statement of a referral by a member of the primary-care team to a third-party exercise provider. Examples of three such studies are presented in Boxes 6–8. A full list of excluded papers is provided in Appendix 3.

BOX 6 - Example I of recruitment from primary care

Hardcastle et al.65

‘Participants were drawn from a patient electronic database at a local health centre.’

‘A total of 1439 patients were contacted by mail with an invitation letter and information sheet telling them about the study. Three hundred and fifty-eight (28%) accepted the invitation to enter the study by completing a form and returning it in a stamped addressed envelope.’

BOX 7 - Example II of recruitment from primary care

Lawton et al.66

‘General practitioners at participating practices were asked to identify women in the age group from their practice register, excluding patients deemed inappropriate for participation in a physical activity trial. The general practitioners sent letters to those identified as suitable, inviting them to participate in a lifestyle study. The invitation letter requested that women contact the research team if they were interested in learning more about the study using the reply slip and prepaid envelope supplied.’

BOX 8 - Example III of recruitment from primary care

Kolt et al.67

‘Two research assistants recruited patients through three primary care practices from different socioeconomic regions of Auckland, New Zealand, from June 2003 to March 2004. The primary care physicians identified and screened all those aged 65 and older on the practice databases (from their files). Those for whom physical activity was not contraindicated and were contactable at the address and telephone number on the practice database (N = 831) were invited to participate in the study via a letter from their primary care physician and follow-up telephone call from the practice where necessary.’

Characteristics of included primary studies

The characteristics of the seven included ERS studies are summarised in Table 6. 26,27,49,60,67–69 These studies included a total of 3030 participants. All studies were RCTs: five undertaken in the UK,27,28,50,61,68 one in Denmark69 and one in Spain. 70 The studies of Jolly et al. 68 and Gusi et al. 70 used cluster allocation (i.e. allocating participants to ERS and control at the ERS provider and general practice level, respectively). The other included studies undertook participant level randomisation. Studies had a median sample size of 347 (range 54–943) and follow-up duration ranged from 2 to 12 months. The GP was the main referrer, usually using a bespoke referral form to a fitness or exercise instructor/officer.

TABLE 6 - Summary of characteristics of included ERS trials

max., maximum; ref., reference; SDT, self-determination theory.

Taylor et al. provided three publications27,71,72 from which data were extracted; for ease of reading, ref. 25 shall be used, with ref. 69 used for the Psychological well-being section. Sorensen et al. provided two publications69,73 from which data were extracted; for ease of reading, ref. 67 shall be used.

Study	No. of GP practices	Date study conducted	RCT design	Overall n	Randomised (n) (ERS/control)	Comparator group description	Follow-up periods
aTaylor et al. 27 UK	3	January to December 1994	Individual	142	97/45	Initial screen No exercise programme	8, 16, 26 and 37 weeks
Stevens et al. 50 UK	1	Not stated		714	363/351	No exercise programme; sent exercise promotion materials	8 months
Harrison et al. 28 UK	46	March 2000 to December 2001		545	275/270	No exercise programme; sent a written information pack	6, 9 and 12 months
Isaacs et al. 61 UK	88	October 1998 to April 2002		943	317/315/311	Initial assessment No exercise programme, advice only or 10-week walking scheme, 2 × 45 minutes/week, 60–80% of heart rate max., group setting	10 weeks, 6 and 12 months
aSorensen et al. 69 Denmark	14	2005–6	Individual	52	28/24	Initial health profile and motivational counselling (45–60 min/session) No exercise programme	4 and 10 months
Gusi et al. 70 Spain	Four	Not stated	Cluster	287	127/160	Best care in general practice, which consisted of routine care and a recommendation of PA	6 months
Jolly et al. 68 UK	Not reported (13 leisure centre sites)	November 2007 to July 2008		347	184/163	ERS plus SDT programme	3 and 6 months

Inclusion and exclusion criteria

Most studies determined their inclusion and exclusion of participants based on criteria of the ERS they were evaluating (Table 7). Four studies27,50,68,70 excluded patients with any form of heart condition. Gusi et al. 70 excluded patients with severe obesity or major depression and Taylor et al. 27 excluded patients with diabetes. All excluded individuals were considered to be at especially high risk [e.g. systolic blood pressure (SBP) of > 200 mmHg, insulin-dependent diabetes].

TABLE 7 - Summary of inclusion and exclusion criteria for included ERS trials

BMI, body mass index; MI, myocardial infarction.

Study	Age range of patients (years)	Inclusion criteria	Exclusion criteria	Inclusion/exclusion criteria determined/evaluated by	No. of participants excluded
Taylor et al. 27 UK	40–70	Smokers, hypertension (140/90 mmHg), overweight (BMI > 25)	SBP > 200 mmHg, history of MI or angina pectoris, diabetes mellitus, musculoskeletal condition preventing PA, previous ERS referral	Research team and GP determined and evaluated	44
Stevens et al. 50 UK	18+	Sedentary–less than 20 × 30 minutes of moderate-intensity PA or less than 12 × 20 vigorous-intensity PA in the last 4 weeks	Medical reasons for exclusion, e.g. registered disabled, diagnosis of heart disease	Research team determined and evaluated	113
Harrison et al. 28 UK	18+	Sedentary, participating in < 90 minutes of moderate/vigorous PA a week, additional CHD risk factors; obesity, previous MI, on practice CHD risk management register, diabetes	GP identified contradiction to PA, SBP > 200 mmHg, not sedentary, only one family member (to avoid contamination–research team criterion)	GP evaluation using the trial’s ERS-determined criteria	285
Isaacs et al. 61 UK	40–74	Not active (no definition reported), raised cholesterol, controlled mild/moderate hypertension, obesity, smoking, diabetes, family history of MI at early age	Pre-existing overt CVD, uncontrolled hypertension, uncontrolled insulin-dependent diabetes, psychiatric or physical conditions preventing PA, conditions requiring specialist programme	GP evaluation using criteria determined by an existing ERS	Not reported
Sorensen et al. 69 Denmark	18+	Patients must meet all criteria: (1) having medically controlled lifestyle diseases or at risk of developing lifestyle diseases; (2) motivated to change lifestyle; (3) believed by the GP to be able to improve health from an increased PA level; and (4) willing to pay 750 Danish krone (€100) for the intervention	Not meeting the inclusion criteria	GP evaluation using the trial’s ERS-determined criteria	Not reported
Gusi et al. 70 Spain	60+	Moderately depressed (6–9 points on the Geriatric Depression Scale), overweight (BMI 25–39.9), capable of walking for more than 25 minutes	Severe obesity, major depression, debilitating medical condition, known unstable cardiac condition, attention or comprehension problems	Research team determined, GP evaluation	32
Jolly et al. 68 UK	18+	People with two or more major risk factors of coronary heart disease: family history of CHD, smoking, raised cholesterol obese (BMI > 30 or BMI > 25 plus one other risk factor) People suffering from well-controlled chronic medical conditions: mild or controlled asthma, chronic bronchitis, controlled diabetes mellitus, mild-to-moderate depression and/or anxiety, people for whom the onset of osteoporosis may be delayed through regular exercise: i.e. post-menopausal women, borderline hypertensive patients with a blood pressure no higher than 160/102 mmHg prior to medication, people exhibiting motivation to change	Angina pectoris, moderate-to-high (or unstable) hypertension ≥ 160/102 mmHg Poorly controlled insulin-dependent diabetes, history of MI within the last 6 months – unless the patient has completed stage III cardiac rehabilitation, established cerebrovascular disease, severe chronic obstructive airways disease, uncontrolled asthma	GP evaluation using the trial’s ERS-determined criteria	Not reported

Trial participants

Studies mainly recruited sedentary, middle-aged white adults who had no medical diagnosis and evidence of at least one lifestyle risk factor, i.e. high blood pressure, raised serum cholesterol, smoking or being overweight (Tables 7 and 8). Studies also included a number of individuals with a medical diagnosis that included diabetes, hypertension, depression, CHD and obesity. However, all included studies reported outcomes aggregated across all participants (see Findings, below). Only Gusi et al. 70 reported a rural population with 66% of participants living in a rural area.

TABLE 8 - Summary of participant characteristics of included ERS trials

BMI, body mass index; MI, mycardial infarction.

Study	Age (mean, years)		Gender (% male)		Ethnicity (%)		Reported diagnosed conditions or risk factors (%)
	Intervention	Control	Intervention	Control	Intervention	Control	Intervention	Comparator
Taylor et al. 27 UK	54.1	54.4	37	38	Not reported	Not reported	Smokers: 43% Overweight: 77% Hypertensive: 46%	Smokers: 40% Overweight: 71% Hypertensive: 58%
Stevens et al. 50 UK	59.1	59.2	40	44	White: 87 Black: 5 Asian: 4 Other: 4	White: 83 Black: 4 Asian: 6 Other: 5	BMI > 25: 46 Smoker: 18	BMI > 25: 42 Smoker: 17
Harrison et al. 28 UK	18–44 = 111 45–59 = 101 > 60 = 63	18–44 = 107 45–59 = 98 > 60 = 65	33	34	White: 71.9	White: 74.1	Smoker: 24.4 At least one CHD risk factor: 75.3	Smoker: 20.7 At least one CHD risk factor: 75.2
Isaacs et al. 61 UK	57.1	Usual care: 57 Walk: 56.9	ERS: 35	32 Walk: 31	White: 75.7 Asian:16.7	(Control/walking) White: 76.5/75.9 Asian: 14/12.2	(Exercise/walking) Raised cholesterol: 24.0 Hypertension: 44.5 Obesity: 65.9 Smoking: 10.4 Type 2 diabetes: 12.3/11.3 Family history of MI: 13.9	(Control/walking) Raised cholesterol: 17.1/21.5 Hypertension: 43.5/46.3 Obesity: 63.5/58.5 Smoking: 8.3/12.2 Diabetes: 15.6/11.3 Family history of MI: 16.2/12.9
Sorensen et al. 69 Denmark	53.9	52.9	43	37	Not reported	Not reported	Metabolic syndrome: 36 Type 2 diabetes: 18 CVD: 32 Other diseases: 14	Metabolic syndrome: 25 Diabetes: 21 Heart disease: 42 Other diseases: 13
Gusi et al. 70 Spain	71	74	0	0	Not reported	Not reported	Overweight (BMI>25): 80 Type 2 diabetes: 39 Moderate depression: 34	Overweight: 86 Type 2 diabetes: 37 Moderate depression: 38
Jolly et al. 68 UK	< 30: 19 30–49: 76 50–64: 64 65+: 25	< 30: 11 30–49: 77 50–64: 50 65+: 25	24	30	White: 74.9 Black: 10.6 Asian: 9.5 Other: 5	White: 67.5 Black: 14.9 Asian: 14.9 Other: 2.6	Smoker: 22.1 Hypertension: 38 Overweight (BMI > 25): 25.3 Obese (BMI > 30): 52.3 Morbidly obese (BMI>40): 12.1 Probable anxiety: 34.2 Probable depression: 21.9	Smoker: 23.1 Hypertensive: 37.5 Overweight: 26.3 Obese: 51.9 Morbidly obese: 13.5 Probable anxiety: 31.9 Probable depression: 15.3

Exercise referral scheme intervention

The ERS intervention of all studies, except that of Gusi et al. ,70 undertook an initial consultation by the third-party provider, such as an exercise professional (Table 9). The consultations varied in content, but all contained information and advice about being physically active. Other components of the screen (dependent on study outcomes) included lifestyle and health questionnaires and physical fitness measures. Scheme length was typically 10–12 weeks, and took place in a leisure centre,27,28,50,61,68 a clinic, public parks or forest tracks (Table 10). Exercise sessions were usually twice per week, lasted between 30 and 60 minutes per session, and were conducted at either a moderate or individually tailored intensity. Two studies69,70 provided group-based exercise sessions, and four27,28,61,68 provided a combination of group and individual exercise sessions. Only three studies28,50,68 reported an assessment at the end of the ERS programme.

TABLE 9 - Summary of referral characteristics of included ERS trials

Study	Referrer	Format of referral	Referred to where	Participant cost	Referred to who
Taylor et al. 27 UK	GP	Signed prescription card	Leisure centre	Half-price admission	Fitness instructor
Stevens et al. 50 UK	GP	Letter	Leisure centre	Not reported	Exercise development officer
Harrison et al. 28 UK	GP	Faxed referral form	Leisure centre	‘Subsidised ’	Exercise officer
Isaacs et al. 61 UK	GP or practice nurse	Specially prepared ‘prescription pad’ – referral form	Leisure centre	Free	Fitness instructor
Sorensen et al. 69 Denmark	GP	Not reported	Clinic	Pay €100	Physiotherapist
Gusi et al. 70 Spain	GP	Not reported	Supervised walks in a public park or forest tracks	Not reported	Qualified exercise leaders
Jolly et al. 68 UK	Member of the primary-care team	Not reported	Leisure centre	Not reported	Health and fitness adviser

TABLE 10 - Summary of ERS intervention characteristics of included ERS trials

Study	Initial screen/assessment	Scheme duration	Exercise programme				Exit assessment
			Provider	Exercise sessions per week	Exercise session intensity	Group or individual
Taylor et al. 27 UK	Yes	10 weeks	Leisure centre	2 × 30–40 minutes	Moderate intensity	Group and/or individual	Not reported
Stevens et al. 50 UK	Yes	10 weeks	Leisure centre	Not reported	Not reported	Not reported	Yes
Harrison et al. 28 UK	Yes	12 weeks	Leisure centre	2 × 1 hour	Individually based	Group and/or individual	Yes
Isaacs et al. 61 UK	Yes	10 weeks	Leisure centre	2 × 45 minutes	Not reported	Group and/or Individual
Sorensen et al. 69 Denmark	Yes (and motivational counselling)	4 months	Clinic	First 2 months 2 sessions × 1 hour Second 2 months 1 session × 1 hour	More than 50% of heart rate reserve for a minimum of 20 minutes	Group
Gusi et al. 70 Spain	Not reported	6 months	Walking scheme	3 × 50 minutes	Not reported	Group
Jolly et al. 68 UK	Yes	12 weeks	Leisure centre	Individually based	Individually based	Group and/or Individual

Control/comparator group

Five studies27,28,50,61,70 compared ERS with a ‘usual care’ control group, which consisted of no exercise intervention or simple advice on PA (see Table 6). Sorensen et al. 69 compared ERS with motivational counselling aimed at increasing daily PA. In addition to a no-exercise group, the Isaacs et al. study61 also included an instructor-led walking programme. The Jolly et al. study68 compared two forms of ERS, i.e. standard ERS versus a combined ERS plus self-determination theory (SDT)-based intervention.

Risk of bias

Table 11 summarises the risk of bias for each of the included studies. Most included a power calculation and allocated participants using an appropriately generated random number sequence. However, the reporting of concealment of trial group allocation was poor, although there was good evidence of participant characteristics of intervention and control groups at baseline. Although blinding of participants and intervention providers in these studies was not feasible, blinding of outcome assessment was possible. Outcome blinding is particularly important in preventing assessment bias in the case of outcomes that require observer judgement or involvement (e.g. blood pressure measurement or exercise testing). However, only the study of Jolly et al. 68 reported outcome blinding, i.e. self-reported PA using the 7-Day Physical Activity Recall Questionnaire was assessed via telephone to maintain blinding. The reporting and handling of missing data was detailed for most studies, and all studies, except one,27 reported the use of intention-to-treat (ITT) analysis. The level of missing data at follow-up ranged across studies from 16.5% to 50%. Most studies used imputation methods (last observation carried forward or complete case average values) to replace missing data values at follow-up. Overall, three studies were judged to be at moderate overall risk of bias27,28,50 and four to be at low overall risk of bias. 61,68–70

TABLE 11 - Summary of risk of bias assessment

+, Correspondence with author.

Risk of bias criterion	Taylor et al.27 UK	Stevens et al.50 UK	Harrison et al.28 UK	Isaacs et al.61 UK	Sorensen et al.69 Denmark	Gusi et al.70 Spain	Jolly et al.68 UK
Power calculation reported?	Yes	Unclear	Yes	Yes	Yes	Yes	Yes
Method of random sequence generation described?	Yes	Yes	Yes	Yes	Yes	Yes	Yes+
Method of allocation concealment described?	Yes+	Unclear	Unclear	Unclear	Yes	Yes	Unclear
Method of outcome (assessment) blinding described?	Unclear	Unclear	Unclear	No	Unclear	Unclear	Yes
Were groups similar at baseline?	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Was ITT analysis used?	No	Yes	Yes	Yes	Yes	Yes	Yes
Was there any statistical handling of missing data?	Unclear	Yes	Unclear	Yes	Yes	Yes	Yes
Were missing data (dropout and loss to follow-up) reported?	Yes	Yes	Yes	Yes	Yes	Yes	Yes

Exercise referral scheme eligibility, uptake and adherence

There was a considerable range in the proportion of individuals randomised compared with those deemed eligible (Table 12). In both the Sorensen et al. 69 and Jolly et al. 68 studies, of those deemed eligible for ERS, a substantial number refused participation in the trial. For Sorensen et al. 69 this low number maybe reflective of the €100 payment by patients as part of a standard Danish EoP.

TABLE 12 - Summary of eligibility and uptake figures for included studies

+, Percentage of individuals deemed eligible who were randomised.

Study	No. deemed eligible (n)	Total n randomised	ERS (n)	Control (n)	ERS uptake
Taylor et al. 27 UK	345	142 (41%)+	97	45	85 (88%)
Stevens et al. 50 UK	827	714 (86%)+	363	351	126 (35%)
Harrison et al. 28 UK	830	545 (66%)+	275	270	232 (84%)
Isaacs et al. 61 UK	1305	949 (73%)+	317	315 + 311	293 (92%)
Sorensen et al. 69 Denmark	327	52 (16%)+	28	24	28 (100%)
Gusi et al. 70 Spain	160	127 (79%)+	64	63	Not reported
Jolly et al. 68 UK	1683	347 (21%)+	184	163	Not reported

Uptake was defined as the proportion of those individuals offered entry to ERS who attended an initial consultation with an ‘exercise professional’ or attended a first exercise session. Although Taylor et al. ,27 Issacs et al. 61 and Sorensen et al. 69 reported uptake rates in excess of 85%, in the Stevens et al. 50 study only 126 (35%) of the 233 randomised to ERS attended the first consultation. Stevens et al. 50 discussed how the low uptake they experienced may have been reflective of the nature of the invitation letter sent to participants and the point of randomisation (pre-invitation letter). Furthermore, they hypothesise that a change in the format of the letter (e.g. including a specific appointment date for the first ERS appointment) would have improved participation. Uptake was not reported by Jolly et al. 68 or Gusi et al. 70

Harrison et al. 28 and Jolly et al. 68 failed to provide information on participants’ adherence to the ERS intervention. Stevens et al. 50 and Gusi et al. 70 reported ERS programme completion rates of 25% and 86%, respectively. However, these rates do not reflect the number of sessions attended, only those who attended a second consultation50 or follow-up assessment. 70

Sorensen et al. 69 reported that an average 18 of a total of 24 ERS exercise sessions were attended and 68% and 75% of participants attended the counselling sessions at 4 and 10 months, respectively. Both Taylor et al. 27 and Isaacs et al. 61 provide a detailed description of ERS programme adherence. Taylor27 reported 13% attending no exercise sessions and 28% attending 75–100% of exercise sessions, with an average of 9.1 out of 20 prescribed exercise sessions attended. Isaacs et al. 61 reported 7.6% attending no exercise sessions and 42% attending 75–100% of exercise sessions in the leisure centre group. In the walking group, 23.5% attended no exercise sessions, with 21.5% attending 75–100% of exercise sessions. As shown in Table 13, there was no consistent difference in attendance rates between those in at-risk groups and the overall study population in the studies of Taylor et al. 27 and Isaacs et al. 61 In the Isaacs et al. 61 study, the 60- to 69-year age group had the highest adherence in both the ERS (53.3%) and the walking (24.2%) groups. There were no significant differences in attendance rate with employment status, educational level, socioeconomic status, ethnicity or relationship status. Adherence was lower for those without access to private transport in both the ERS and walking groups.

TABLE 13 - Proportion of individuals by risk group with 75–100% ERS attendance rates

Study	Smoking (%)	Obesity (%)	Hypertension (%)	Overall (%)
Taylor et al. 27 UK	12	28	23	28
Isaacs et al. ERS group61 UK	45.5	38.8	46.1	42
Isaacs et al. control walking group61 UK	26.3	18.7	22.9	21.5

Findings

Only Isaacs et al. 61 reported all outcome domains applicable to this systematic review (Table 14). Outcome results are reported according to the three categories of comparator, i.e. ERS versus usual care; ERS versus alternative exercise intervention and ERS versus alternative form of ERS.

TABLE 14 - Summary of outcome domains assessed

BF%, body fat %; BMI, body mass index; BP, blood pressure; EQ-5D, European Quality of Life-5 Dimensions; FEV, forced expiratory volume; HR, heart rate; LTPAQ, Leisure Time Physical Activity Questionnaire; N/A, not applicable; PAR, Physical Activity Recall; PEF, peak expiratory flow; PSW, physical self-worth; QoL, quality of life; SF-12, Short Form questionnaire-12 items; sub-max, sub-maximal.

Study	PA	PA measure	Physical fitness	Clinical outcomes	Psychological well-being	HRQoL	Patient satisfaction	Adverse events
Taylor et al. 27 UK	Yes	Self-report 7-day PAR	Yes Sub-max HR	Yes BP, BMI, BF%, waist to hip	Yes PSW	No	No	No
Stevens et al. 50 UK	Yes	Self-report 7-day PAR	No	No	No	No	No	No
Harrison et al. 28 UK	Yes	Self-report 7-day PAR	No	No	No	No	Yes	No
Isaacs et al. 61 UK	Yes	Self-report Minnesota LTPAQ	Yes Sub-max bike test Sub-max walking test	Yes BP, cholesterol, lipoproteins, triglycerides, weight, BMI, BF%, waist-to-hip ratio, FEV, PEF	Yes Anxiety, depression	Yes SF-36 mental	Yes	Yes GP records
Sorensen et al.69 Denmark	Yes	Self-report Unspecified	Yes Sub-max bike test	Yes Weight, BMI	No	Yes SF-12 mental, physical	No	No
Gusi et al. 70 Spain	Not reported	N/A	No	Yes BMI	Yes Anxiety, depression	Yes EQ-5D	No	No
Jolly et al. 68 UK	Yes	Self-report 7-day PAR	No	Yes BMI	Yes Anxiety, depression	Yes Dartmouth QoL	No	No

Physical activity

All studies, with the exception of Gusi et al. ,70 provided a measure of self-reported PA. Self-reported measures included the validated 7-Day Physical Activity Recall Questionnaire,27,28,50,68 a modified version of the validated Minnesota Leisure Time Activity Questionnaire61 and an invalidated questionnaire designed by the research team. 63 No studies reported assessed PA using objective methods. A summary of the main PA outcomes at follow-up is provided in Table 15.

TABLE 15 - Summary of PA data at follow-up

SD, standard deviation.

Sorensen et al. 69 metabolic equivalent (METS)/hour/day.

Numbers of individuals with complete data/questionnaires.

Between-group difference not statistically significant at p ≤ 0.05.

The p-value calculated by authors of the present report.

Between-group difference statistically significant at p ≤ 0.05.

All randomised participants.

Provides b (patients achieving PA guidance) and f (all other PA measures).

Mean change score.

Study and time of follow-up	Patients achieving PA guidance (90–150 minutes/at least moderate-intensity per week)		Minutes per week at least moderate intensity				Total PA (minutes per week)		Energy expenditure (kcal/kg/day)a
	ERS, n/N	Usual care, n/N	ERS, mean (SD)		Usual care, mean (SD)		ERS, mean (SD)	Usual care, mean (SD)	ERS, mean (SD)	Usual care, mean (SD)
ERS vs usual care
Taylor et al.27
			(Moderate)	(Vigorous)	(Moderate)	(Vigorous)
8 weeksb	51/63	20/31c,d	247 (174)	49 (60)	145 (178)e	21 (61)e	Not reported	Not reported	34.6 (1.2)	33.7 (1.7)e
16 weeksb	51/57	18/31d,e	226 (252)	59 (72)	160 (262)c	21 (72)e	Not reported	Not reported	34.6 (1.2)	33.9 (1.7)e
26 weeksb	39/47	18/31d,e	183 (234)	56 (108)	206 (251)c	34 (111)c	Not reported	Not reported	34.4 (1.8)	34.3 (1.2)c
37 weeksb	39/57	19/c,d	158 (228)	42 (96)	162 (245)c	23 (106)c	Not reported	Not reported	34.1 (2.4)	33.9 (2.2)c
Stevens et al.50
8 monthsf	204/363	174/351c,d	Not reported		Not reported		Not reported	Not reported	Not reported	Not reported
Harrison et al.28
6 monthsb	38/168	22/162d,e	Not reported		Not reported		Not reported	Not reported	Not reported	Not reported
9 monthsb	36/149	31/140c	Not reported		Not reported		Not reported	Not reported	Not reported	Not reported
12 monthsb	40/155	32/157c	Not reported		Not reported		Not reported	Not reported	Not reported	Not reported
Isaacs et al.61
10 weeksg	48/164	29/157d,e	93 (115)		79 (114)c		584 (479)	668 (555)c	34 (26)	36 (32)c
6 monthsg	70/179	66/200c,d	65 (106)		58 (98)c		692 (496)	647 (463)c	38 (27)	35 (27)c
Gusi et al.70
	Not reported	Not reported	Not reported		Not reported		Not reported	Not reported	Not reported	Not reported
Study and time of follow-up	Patients achieving PA guidance (90–150 minutes/at least moderate-intensity per week)		Minutes per week at least moderate intensity				Total PA (minutes per week)		Energy expenditure (kcal/kg/day)a
	ERS, n/N	Alternative PA, n/N	ERS, mean (SD)		Alternative PA, mean (SD)		ERS, mean (SD)	Alternative PA, mean (SD)	ERS, mean (SD)	Alternative PA, mean (SD)
ERS vs alternative PA intervention
hSorensen et al.69
4 monthsb	Not reported	Not reported	Not reported		Not reported		63 (114)	23 (107)c	43 (2.4)	41 (4.8)
10 monthsb	Not reported	Not reported	Not reported		Not reported		20 (124)	20 (152)c	41 (2.1)	40 (5)
Isaacs et al.61
10 weeksg	48/164	53/92d,e	93 (115)		113 (291)c		584 (479)	863 (1026)e	34 (26)	43 (38)e
6 monthsg	70/179	62//141c,d	65 (106)		89 (150)e		692 (496)	759 (539)c	38 (27)	42 (27)c
	ERS, n/N	ERS plus SDT, n/N	ERS, mean (SD)		ERS plus SDT, mean (SD)		ERS, mean (SD)	ERS plus SDT, mean (SD)	ERS, mean (SD)	ERS plus SDT, mean (SD)
ERS vs ERS plus SDT
Jolly et al.68
3 monthsg	Not reported	Not reported	319 (338)c		331 (336)c		Not reported	Not reported	Not reported	Not reported
6 monthsg	66/156	83/169c,d	249 (356)c		246 (343)c		Not reported	Not reported	Not reported	Not reported

Exercise referral scheme versus usual care

The most consistently reported PA outcome across studies was the proportion of individuals achieving 90–150 minutes of at least moderate-intensity activity per week. (The use of 90–150 minutes of at least moderate-intensity PA/week is pragmatic with the included studies.) When pooled across studies there was a 16% (95% CI 3% to 30%) increase in the RR of achieving this outcome with ERS compared with usual care at 6–12 months’ follow-up (Figure 2).

FIGURE 2.

Meta-analysis of the number of patients achieving 90–150 minutes of at least moderate-intensity PA per week at 6–12 months’ follow-up (denominators as reported by authors).

The studies of Taylor et al. 27 and Harrison et al. 28 reported this outcome based on the number of individuals who were available at follow-up. In order to assess the potential (attrition) bias in using completers, we adjusted the denominators of these two studies to all individuals randomised – an ITT analysis (Figure 3). We assumed that all missing cases did not meet the PA threshold. In the pooled ITT analysis, the proportion achieving the PA threshold with ERS than usual care (11%, 95% CI –1% to 45%) this effect was no longer statistically significant.

FIGURE 3.

Meta-analysis of the number of patients achieving 90–150 minutes of at least moderate-intensity PA per week at 6–12 months’ follow-up (denominators adjusted to all randomised – ITT analysis).

There was no difference between ERS and usual care in either the minutes spent in at least moderate-intensity PA/week or estimated PA-induced energy expenditure (Figures 4 and 5).

FIGURE 4.

Meta-analysis of patient’s minutes spent in at least moderate-intensity PA per week at 6–12 months’ follow-up. SD, standard deviation.

FIGURE 5.

Meta-analysis of energy expenditure (kcal/kg/day) ERS vs usual care at 6–12 months’ follow-up. SD, standard deviation.

Exercise referral scheme versus alternative physical activity intervention

Sorensen et al. 69 reported a higher level of energy expenditure with ERS than with PA counselling. In contrast, the study by Isaacs et al. 61 observed a higher level of PA (minutes of total and moderate-intensity activity, and energy expenditure) in those in the walking programme than in the ERS group. When pooled across studies, there was no significant difference in the total amount of physical or energy expenditure between ERS and alternative PA interventions (Figures 6 and 7).

FIGURE 6.

Meta-analysis of patient’s minutes of total PA/week at 6–12 months’ follow-up. SD, standard deviation.

FIGURE 7.

Meta-analysis of energy expenditure ERS vs alternative PA intervention at 6–12 months’ follow-up. SD, standard deviation.

Exercise referral scheme versus exercise referral scheme plus self-determination theory

In the Jolly et al. study,68 the proportion of patients achieving at least 150 minutes of moderate PA per week increased in the standard ERS group from 27% at baseline to 63% at 3 months and 46% at 6 months. There were no significant differences in these proportions between the standard ERS and ERS-plus-SDT groups (Table 15).

Subgroup analysis

Harrison et al. 28 reported no statistically significant interaction effects between the ERS effect and pre-specified baseline variables (i.e. CHD risk factors, sex and age). Comparing high adherers (≥ 75% attendance at ERS) with low adherers (< 75% attendance at ERS) in the Isaacs et al. study,61 32 high adherers and 16 low adherers were achieving ≥ 150 minutes of moderate PA per week at 10 weeks. At 6 months, 41 high adherers and 29 low adherers were achieving ≥ 150 minutes of moderate PA per week. However, these proportions were not significantly different. In the Jolly et al. study,68 age, gender, deprivation (Index of Multiple Deprivation score), ethnicity, depression at baseline and level of PA at baseline were assessed by regression methods as predictors of PA at 6 months. Only PA at baseline was associated with PA at the 6-month follow-up (p < 0.001).

Physical fitness

The studies by Taylor et al. ,27 Isaacs et al. 61 and Sorensen et al. 69 reported physical fitness outcomes (Table 16).

TABLE 16 - Summary of physical fitness data at follow-up in included ERS trials

N, newton; N/A: not applicable; SD, standard deviation; VO_2max, maximal oxygen uptake; W, watt.

Numbers of individuals with complete data/questionnaires.

Between-group difference not statistically significant at p ≤ 0.05.

Between-group differences statistically significant at p ≤ 0.05.

Study and time of follow-up	Mean predicted heart rate at a workload of 150 W		VO2max (ml/kg/minute)		Submaximal bike ergometer (minutes)		Submaximal shuttle walk (m)		Isometric knee strength (N)		Leg extension power (W)
	ERS, mean (SD)	Usual care, mean (SD)	ERS, mean (SD)	Usual care, mean (SD)	ERS, mean (SD)	Usual care, mean (SD)	ERS, mean (SD)	Usual care, mean (SD)	ERS, mean (SD)	Usual care, mean (SD)	ERS, mean (SD)	Usual care, mean (SD)
ERS vs usual care
Taylor et al.27
16 weeksa	138.6 (23.0)	147.2 (29.7)b	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported
26 weeksa	136.3 (22.6)	142.3 (28.5)b	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported
37 weeksa	134.2 (19.0)	146.0 (24.2)c	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported
Isaacs et al.61
10 weeksa	Not reported	Not reported	Not reported	Not reported	9.65 (1.5)	8.87 (1.5)c	456 (102)	434 (104)b	277 (54)	265 (56)b	174 (31)	165 (31)b
6 monthsa	Not reported	Not reported	Not reported	Not reported	8.86 (1.7)	9.08 (1.7)b	445 (96)	434 (97)b	265 (58)	267 (66)b	173 (66)	167 (68)b
	ERS, mean (SD)	Alternative PA, mean (SD)	ERS, mean (SD)	Alternative PA, mean (SD)	ERS, mean (SD)	Alternative PA, mean (SD)	ERS, mean (SD)	Alternative PA, mean (SD)	ERS, mean (SD)	Alternative PA, mean (SD)	ERS, mean (SD)	Alternative PA, mean (SD)
ERS vs alternative: PA intervention
Sorensen et al.69
4 monthsa	Not reported	Not reported	23.8 (7.1)	21.7 (11.0)b	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported
10 monthsa	Not reported	Not reported	23.0 (8.2)	22.4 (12.7)b	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported
Isaacs et al.61
10 weeksa	Not reported	Not reported	Not reported	Not reported	9.65 (1.5)	8.92 (1.7)b	456 (102)	437 (100)b	277 (54)	275 (58)b	174 (31)	166 (32)b
6 monthsa	Not reported	Not reported	Not reported	Not reported	8.86 (1.7)	8.92 (1.8)b	445 (96)	448 (95)b	265 (58)	264 (66)b	173 (66)	164 (68)b

Exercise referral scheme versus usual care

Taylor et al. 27 reported a lower (more favourable) submaximal heart rate (at 150 W) for ERS compared with usual care. Isaacs et al. 61 reported no significant differences in any of the physical fitness measures (submaximal bike and shuttle walk, isometric knee strength, leg extension power) between the ERS and usual care groups at follow-up except at 10 weeks for the submaximal bike ergometer test. Pooling of the cardiorespiratory measures (mode: cycle ergometer or cycle/walking) showed no difference between ERS and usual care (Figure 8). There was considerable evidence of statistical heterogeneity.

FIGURE 8.

Meta-analysis of patient’s physical fitness, at 6–12 months’ follow-up. SD, standard deviation.

Exercise referral scheme versus alternative physical activity intervention

Isaacs et al. 61 and Sorensen et al. 69 reported no significant differences in any of the physical fitness measures between the ERS and the alternative PA intervention groups at follow-up (see Figure 8).

Exercise referral scheme versus exercise referral scheme plus self-determination theory

The study of Jolly et al. 68 did not assess physical fitness.

Clinical factors

Five studies27,61,68–70 provided information on clinical outcomes, i.e. CHD risk factors (Table 17), weight and obesity measures (Table 18) and respiratory function (Table 19).

TABLE 17 - Summary of CHD risk factors in included ERS trials

DBP, diastolic blood pressure; N/A, not applicable; SD, standard deviation.

Numbers of individuals with complete data.

Between-group difference not statistically significant at p ≤ 0.05.

Study and time of follow-up	SBP (mmHg)		DBP (mmHg)		Cholesterol (mmol/l)		High-density lipoproteins (mmol/l)		Low-density lipoproteins (mmol/l)		Triglycerides (mmol/l)
	ERS, mean (SD)	Usual care, mean (SD)	ERS, mean (SD)	Usual care, mean (SD)	ERS, mean (SD)	Usual care, mean (SD)	ERS, mean (SD)	Usual care, mean (SD)	ERS, mean (SD)	Usual care, mean (SD)	ERS, mean (SD)	Usual care, mean (SD)
ERS vs usual care
Taylor et al.27
16 weeksa	130 (14.5)	130 (14)b	84 (8)	84 (8)b	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported
26 weeksa	130 (14)	131 (14)b	84 (8)	84 (8)b	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported
37 weeksa	130 (17)	131 (18)b	85 (9)	83 (9)b	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported
Isaacs et al.61
10 weeks	133 (10)	132 (10)b	82 (6)	83 (6)b	5.68 (0.53)	5.71 (0.42)b	1.35 (0.18)	1.35 (0.18)b	3.41 (0.46)	3.44 (0.47)b	2.12 (0.71)	2.14 (0.71)b
6 months	133 (12)	133 (12)b	82 (6)	82 (7)b	5.65 (0.50)	5.60 (0.50)b	1.37 (0.25)	1.38 (0.17)b	3.40 (0.48)	3.37 (0.50)b	2.04 (0.74)	2.00 (0.84)b
	ERS, mean (SD)	Alternative PA, mean (SD)	ERS, mean (SD)	Alternative PA, mean (SD)	ERS, mean (SD)	Alternative PA, mean (SD)	ERS, mean (SD)	Alternative PA, mean (SD)	ERS, mean (SD)	Alternative PA, mean (SD)	ERS, mean (SD)	Alternative PA, mean (SD)
ERS vs alternative PA intervention
Isaacs et al.61
10 weeks	133 (10)	134 (10)b	82 (6)	84 (6)b	5.68 (0.53)	5.69 (0.53)b	1.35 (0.18)	1.33 (0.17)b	3.41 (0.46)	3.45 (0.46)b	2.12 (0.71)	2.05 (0.76)b
6 months	133 (12)	134 (12)b	82 (6)	83 (6)b	5.65 (0.50)	5.56 (0.57)b	1.37 (0.25)	1.37 (0.16)b	3.40 (0.48)	3.36 (0.48)b	2.04 (0.74)	1.95 (0.74)b
	ERS, mean (SD)	ERS plus SDT, mean (SD)	ERS, mean (SD)	ERS plus SDT, mean (SD)	ERS, mean (SD)	ERS plus SDT, mean (SD)	ERS, mean (SD)	ERS plus SDT, mean (SD)	ERS, mean (SD)	ERS plus SDT, mean (SD)	ERS, mean (SD)	ERS plus SDT, mean (SD)
ERS vs ERS plus SDT
Jolly et al.68
6 monthsa	130 (17)	127 (16)b	82 (11)	79 (11)b	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported

TABLE 18 - Summary of weight and measures of obesity outcomes in included ERS trials

N/A, not applicable; SD, standard deviation.

Taylor et al. ,27 sum of four skinfolds (mm).

Numbers of individuals with complete data.

Between-group difference not statistically significant at p ≤ 0.05.

Between-group differences statistically significant at p ≤ 0.05.

All randomised participants.

Mean change score.

Study and time of follow-up	Weight (kg)		BMI (kg/m2)		Body fat (%)a		Waist–hip ratio (cm)
	ERS, mean (SD)	ERS, mean (SD)	ERS, mean (SD)	Usual care, mean (SD)	ERS, mean (SD)	Usual care, mean (SD)	ERS, mean (SD)	Usual care, mean (SD)
ERS vs usual care
Taylor et al.27
16 weeksb	Not reported	Not reported	27.5 (0.6)	27.6 (0.6)c	70 (8)	76 (8)d	0.87 (0.08)	0.83 (0.09)c
26 weeksb	Not reported	Not reported	27.3 (1.3)	27.5 (1.1)c	70 (11)	75 (11)d	0.87 (0.08)	0.83 (0.09)c
37 weeksb	Not reported	Not reported	27.5 (1.3)	27.6 (1.1)c	71 (13)	76 (13)d	0.87 (0.08)	0.84 (0.09)c
Isaacs et al.61
10 weekse	81 (3)	81 (3)c	30.2 (0.8)	30.1 (1.5)c	37.4 (1.9)	37.5 (1.9)c	0.88 (0.06)	0.89 (0)c
6 monthse	82 (3)	82 (3)c	30.5 (1.1)	30.4 (1.1)c	37.8 (2.4)	37.8 (2.4)c	0.88 (0)	0.88 (0)c
Gusi et al.70
6 monthsb	Not reported	Not reported	29.7 (4.2)	30.6 (4.3)d	Not reported	Not reported	Not reported	Not reported
Study and time of follow-up	Weight (kg)		BMI (kg/m2)		Body fat (%)a		Waist–hip ratio (cm)
	ERS, mean (SD)	ERS, mean (SD)	ERS, mean (SD)	Alternative PA, mean (SD)	ERS, mean (SD)	Alternative PA, mean (SD)	ERS, mean (SD)	Alternative PA, mean (SD)
ERS vs alternative PA intervention
fSorensen et al.69
4 monthsb	–1.1 (4)	–1.1 (4)c	–0.3 (1.3)	–0.04 (1.6)c	Not reported	Not reported	Not reported	Not reported
10 monthsb	–0.3 (4.4)	–0.3 (4.4)c	–0.1 (1.9)	–0.6 (2.8)c	Not reported	Not reported	Not reported	Not reported
Isaacs et al.61
10 weekse	81 (3)	81 (3)c	30.2 (0.8)	30.2 (1.6)c	37.4 (1.9)	37.1 (1.9)c	0.88 (0.06)	0.88 (0.06)c
6 monthse	82 (3)	82 (3)c	30.5 (1.1)	30.5 (1.1)c	37.8 (2.4)	37.8 (1.1)c	0.88 (0)	0.88 (0)c
	ERS, mean (SD)	ERS plus SDT, mean (SD)	ERS, mean (SD)	ERS plus SDT, mean (SD)	ERS, mean (SD)	ERS plus SDT, mean (SD)	ERS, mean (SD)	ERS plus SDT, mean (SD)
ERS vs ERS plus SDT
Jolly et al.68
6 monthse	Not reported	Not reported	32.8 (6.9)	32.8 (6.4)c	Not reported	Not reported	Not reported	Not reported

TABLE 19 - Summary of respiratory function outcomes in included ERS trials

FEV, forced expiratory volume; FVC, forced vital capacity; PEF, peak expiratory flow; SD, standard deviation.

All randomised participants.

Between-group difference not statistically significant at p ≤ 0.05.

Study and time of follow-up	FEV/FVC ratio		PEF
	ERS, mean (SD)	Usual care, mean (SD)	ERS, mean (SD)	Usual care, mean (SD)
ERS vs usual care
Isaacs et al.61
10 weeksa	0.86 (0.0)	0.86 (0.06)b	417 (58)	409 (58)b
6 monthsa	0.86 (0.09)	0.86 (0.09)b	407 (115)	411 (117)b
	ERS, mean (SD)	Alternative PA, mean (SD)	ERS, mean (SD)	Alternative PA, mean (SD)
ERS vs alternative PA intervention
Isaacs et al.61
10 weeksa	0.86 (0.0)	0.85 (0.06)b	417 (58)	407 (61)b
6 monthsa	0.86 (0.09)	0.85 (0.09)b	407 (115)	416 (117)b

Exercise referral scheme versus usual care

Taylor et al. 27 reported percentage of body fat in ERS participants compared with usual care at follow-up. Gusi et al. 70 reported a lower BMI, with no other between-group differences in weight and body fat outcomes for the other measured clinical factors (Figures 9 and 10). There was no significant difference in resting blood pressure, serum lipids or respiratory function between ERS and usual care at follow-up (Figures 11 and 12).

FIGURE 9.

Meta-analysis of SBP at 6–12 months’ follow-up. SD, standard deviation.

FIGURE 10.

Meta-analysis of diastolic blood pressure at 6–12 months’ follow-up. SD, standard deviation.

FIGURE 11.

Meta-analysis of BMI at 6–12 months’ follow-up. SD, standard deviation.

FIGURE 12.

Meta-analysis of body fat at 6–12 months’ follow-up. SD, standard deviation.

Exercise referral scheme versus alternative physical activity intervention

In both the studies by Isaacs et al. 61 and Sorensen et al. 69 there were no significant between-group differences at follow-up in resting blood pressure (Figures 9 and 10), BMI (Figure 9), body fat outcomes, serum lipids and respiratory function. The Sorensen et al. 69 trial reported reduced levels of glycosylated haemoglobin (HbA_1c) in both the ERS group (mean –0.26%, 95% CI –0.79% to 0.27%) and the PA counselling group (mean –0.23, 95% CI –0.47 to 0.02) at 4-month follow-up, although there was no difference between groups.

Exercise referral scheme versus exercise referral scheme plus self-determination theory

Jolly et al. 68 reported no significant difference between standard ERS and ERS plus SDT in body mass index (BMI) or resting blood pressure.

Psychological well-being

Four studies61,68,70,71 reported psychological well-being outcomes and are summarised in Table 20.

TABLE 20 - Summary of psychological well-being data at follow-up in included ERS trials

SD, standard deviation.

Significant difference in change from baseline between groups.

All randomised participants.

Between-group differences statistically significant at p ≤ 0.05.

Only mean values available.

Numbers of individuals with complete data/questionnaires.

Between-group difference not statistically significant at p ≤ 0.05.

Study	PSW		Anxiety		Depression		Anxiety/depression
	ERS, mean (SD)	Usual care, mean (SD)	ERS, mean (SD)	Usual care, mean (SD)	ERS, mean (SD)	Usual care, mean (SD)	ERS, mean (SD)	Usual care, mean (SD)
ERS vs usual care
aTaylor and Fox71
16 weeksb	2.31 (0.79)	2.31 (0.67)c	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported
37 weeksb	2.41 (0.79)	2.42 (0.54)c	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported
dIsaacs et al.61
6 monthse	Not reported	Not reported	6.9	7.1f	4.8	4.9f	Not reported	Not reported
Gusi et al.70
6 monthse	Not reported	Not reported	14.1 (9)	22.2 (9.8)c	1.8 (2.3)	2.9 (2.5)c	1.2 (0.4)	1.5 (0.7)c
	ERS, mean (SD)	Alternative PA, mean (SD)	ERS, mean (SD)	Alternative PA, mean (SD)	ERS, mean (SD)	Alternative PA, mean (SD)	ERS, mean (SD)	Alternative PA, mean (SD)
ERS vs alternative PA intervention
dIsaacs et al.61
6 monthse	Not reported	Not reported	6.9	7.5f	4.8	5.1f	Not reported	Not reported
	ERS, mean (SD)	ERS plus SDT, mean (SD)	ERS, mean (SD)	ERS plus SDT, mean (SD)	ERS, mean (SD)	ERS plus SDT, mean (SD)	ERS, mean (SD)	ERS plus SDT, mean (SD)
ERS vs ERS plus SDT
Jolly et al.68
3 monthsb	Not reported	Not reported	7.7 (4.4)f	8.89 (4.3)	5.9 (4.2)f	6.68 (4.1)	Not reported	Not reported
6 monthsb	Not reported	Not reported	7.9 (4.8)f	8.86 (4.7)	6.1 (4.4)f	6.65 (4.3)	Not reported	Not reported

Exercise referral scheme versus usual care

Taylor and Fox71 reported physical self-perceptions measures, with improvements shown in physical self-worth (PSW), and perceptions of physical condition and physical health collected physical self-perceptions data, and reported significant in the ERS group compared with usual-care group at 16 and 37 weeks. Isaacs et al. 61 reported no differences between the ERS and usual-care groups in the anxiety and depression scores using the Hospital Anxiety Depression Scale (HADS) at 6 months. In the Gusi et al. 70 study, all measures [Geriatric Depression Scale, State Trait Anxiety Inventory and the anxiety/depression subscale of the European Quality of Life-5 Dimensions (EQ-5D)] at 6 months were found to favour ERS participants compared with those receiving the usual care.

Exercise referral scheme versus alternative physical activity intervention

Isaacs et al. 61 reported no differences between the ERS and walking programme in anxiety or depression outcomes at 6 months’ follow-up.

Exercise referral scheme versus exercise referral scheme plus self-determination theory

Jolly et al. 68 reported no difference between groups in anxiety or depression outcomes at either 3 or 6 months’ follow-up.

Health-related quality of life

Four studies61,68–70 reported HRQoL, as summarised in Table 21.

TABLE 21 - Summary of HRQoL data at follow-up in included ERS trials

QoL, quality of life; SD, standard deviation.

Only mean values available.

All randomised participants.

Between-group difference not statistically significant at p ≤ 0.05.

Numbers of individuals with complete data/questionnaires.

Between-group differences statistically significant at p ≤ 0.05 (p-value calculated by authors of the present report).

Study and time of follow-up	SF-36 mental		SF-12 mental		SF-12 physical		EQ-5D		Dartmouth QoL (overall QoL scale)
	ERS, mean (SD)	Usual care, mean (SD)	ERS, mean (SD)	Usual care, mean (SD)	ERS, mean (SD)	Usual care, mean (SD)	ERS, mean (SD)	Usual care, mean (SD)	ERS, mean (SD)	Usual care, mean (SD)
ERS vs usual care
aIsaacs et al.61
6 monthsb	54.2	54.3c	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported
Gusi et al.70
6 monthsd	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	0.89 (0.18)	0.51 (0.2)e	Not reported	Not reported
	ERS, mean (SD)	Alternative PA, mean (SD)	ERS, mean (SD)	Alternative PA, mean (SD)	ERS, mean (SD)	Alternative PA, mean (SD)	ERS, mean (SD)	Alternative PA, mean (SD)	ERS, mean (SD)	Alternative PA, mean (SD)
ERS vs alternative PA intervention
Sorensen et al.69
4 monthsb	Not reported	Not reported	40 (10.7)	37 (11.9)c	49 (1017.6)	46 (13.1)c	Not reported	Not reported	Not reported	Not reported
10 monthsb	Not reported	Not reported	41 (10.8)	39 (10.9)c	51 (11.6)	45 (15.4)c	Not reported	Not reported	Not reported	Not reported
aIsaacs et al.61
6 monthsb	54.3	53c	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported
	ERS, mean (SD)	ERS plus SDT, mean (SD)	ERS, mean (SD)	ERS plus SDT, mean (SD)	ERS, mean (SD)	ERS plus SDT, mean (SD)	ERS, mean (SD)	ERS plus SDT, mean (SD)	ERS, mean (SD)	ERS plus SDT, mean (SD)
ERS vs ERS plus SDT
Jolly et al.68
3 months	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	3.16 (0.8)c	3.25 (0.7)c
6 months	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	Not reported	3.15 (0.8)c	3.24 (0.8)c

Exercise referral scheme versus usual care

Isaacs et al. 61 reported no differences between the ERS and usual-care groups at follow-up on the Short Form questionnaire-36 items (SF-36) mental health scale. Gusi et al. 70 observed higher EQ-5D scores in the ERS group than in the usual care group at 6 months.

Exercise referral scheme versus alternative physical activity intervention

Isaacs61 reported no differences between the ERS and walking groups at follow-up on the SF-36 mental health scale score. Similarly, Sorensen69 found no differences between the groups at follow-up on the Short Form questionnaire-12 items (SF-12) mental and physical scales.

Exercise referral scheme vs exercise referral scheme plus self-determination theory

Jolly et al. 68 reported no difference between groups in overall Dartmouth CO-OP chart score although there was a difference for the feelings subscale at 6 months in favour of the alternative ERS group (not tabularised).

Patient satisfaction

Three studies27,28,61 reported patient satisfaction and results are summarised in Table 22.

TABLE 22 - Summary of participant satisfaction in included ERS trials

Statistically significant at p ≤ 0.05 (p-value calculated by authors of the present report).

Difference not statistically significant at p ≤ 0.05.

Study	Satisfied with received information (%)		Needed further information (%)
	ERS	Usual care	ERS	Usual care
ERS vs usual care
Harrison et al.28
3 months	92	69a	43	54a
	ERS	Alternative PA	ERS	Alternative PA
ERS vs alternative PA intervention
Isaacs et al.61
10 weeks	97	96b	15	17b

Exercise referral scheme versus usual care

The Harrison et al. study28 reported that the ERS group were significantly more satisfied with the information they received and felt they needed less information about PA, compared with usual care group. In the Taylor et al. 27 study, comments about the concept of ERS (measured at 8 weeks) identified that 50% of patients were positive, 35% had mixed feelings and 15% had only negative comments. Negative comments included a long waiting time before introductory session, lack of staff support, crowded facilities and inconvenient facility times.

Exercise referral scheme versus alternative physical activity intervention

In the Isaacs et al. 61 study there was no between-group difference in participant satisfaction with received information or the need for additional information. In the ERS group, 97.8% felt better for taking part and enjoyed the programme compared with 93.8% feeling better for taking part and 95.2% enjoying the programme for the walking group.

Exercise referral scheme versus exercise referral scheme plus self-determination theory

Jolly et al. 68 did not assess participant satisfaction.

Adverse events

Although participation in ERS has the potential to lead to negative events (e.g. an increase in exercise-related musculoskeletal injuries or exercise-related cardiac complications), only the Isaacs et al. 61 study assessed such events. Using GP records, the authors assessed the change in consultations before and after ERS. There was evidence of a small increase in GP visits for falls and fractures in the ERS and walking groups compared with usual care control after the start of the study (Table 23).

TABLE 23 - Adverse events reported by the Isaacs et al.61 UK study (GP visits)

Adverse events	Leisure centre	Walking control	Advice-only control
Visits for chest pain
12–6 months before start of study	1 (%)	3	7
6 months before start of study	3 (%)	4	7
Start of study to 6 months	2 (%)	9	7
6–12 months after start of study	10 (%)	4	–
Visits for aches/pains
12–6 months before start of study	54	48	56
6 months before start of study	62	53	55
Start of study to 6 months	52	42	44
6–12 months after start of study	63	44	–
Visits for sprains
12–6 months before start of study	2	2	7
6 months before start of study	3	6	2
Start of study to 6 months	1	4	6
6–12 months after start of study	2	0	–
Visits for falls
12–6 months before start of study	1	1	0
6 months before start of study	1	1	2
Start of study to 6 months	9	2	0
6–12 months after start of study	3	6	–
Visits for fractures
12–6 months before start of study	0	1	1
6 months before start of study	0	0	0
Start of study to 6 months	1	0	0
6–12 months after start of study	0	4	–

Health-care utilisation

No studies reported hospitalisations, primary care visits or use of medication.

Summary

• Given the lack of standardisation of the ERS definition used by previous systematic reviews and the publication of further recent evidence, we undertook a de novo systematic review of the effectiveness of ERS.

• We undertook a search of electronic databases – MEDLINE (Ovid) 1990 to October 2009; EMBASE (Ovid) 1990 to October 2009; PsycINFO; The Cochrane Library (Wiley) 2009 v3 (CENTRAL, DARE, NHS HTA, NHS EED, HTA database), ISI Web of Knowledge (WOK); and SPORTDiscus; ongoing trials registry – and contacted experts in the field to identify unpublished studies. We limited our inclusion criteria to controlled studies (randomised or non-randomised) that met our ERS definition, i.e. (1) referral by a primary-care health-care professional to a third party, (2) that provided a PA programme tailored to individual needs and (3) an initial assessment and monitoring throughout the programme.

• Our systematic review identified seven RCTs (3030 participants: UK, n = 5; non-UK, n = 2). These studies and were heterogeneous in their population, interventions and comparators. Five studies compared ERS to usual care (e.g. PA advice), two compared ERS with an alternative PA-promoting strategy (i.e. walking programme or PA counselling) and one study compared traditional ERS with combined ERS plus SDT intervention. Although all studies recruited predominantly sedentary middle-aged adults who had at least one lifestyle risk factor (i.e. hypertension, raised serum cholesterol, smoking or being overweight), a number of the studies also included a proportion of specific medical diagnoses [i.e. myocardial infarction (MI), type 2 diabetes, obesity (BMI > 35 kg/m²), hypertension and depression]. ERS mainly took place at a leisure centre and typically involved 10–12 weeks of exercise intervention, with the longest study reporting outcomes up to 6–12 months post baseline measures. Uptake (proportion of individuals randomised to ERS who attended the first exercise session) varied widely across studies (35–85%), as did adherence to the ERS intervention (programme completion rates of 25–86%).

• Studies were judged to have a low to moderate overall risk of bias. Outcome blinding for PA interventions of this nature is difficult to implement, with other quality issues generally poorly reported as opposed to not being implemented.

• The most consistently reported outcome was self-reported PA. Pooling across four studies, compared with usual care, 11% (95% CI –2% to 26%) more ERS participants achieved 90–150 minutes of at least moderate-intensity PA per week at 6–12 months’ follow-up (ITT analysis). There was no significant difference in PA between ERS versus alternative PA promotion intervention or ERS versus ERS plus SDT at 6–12 months’ follow-up. Other reported measures of PA (i.e. amount of total and moderate PA and energy expenditure) did not show a difference between ERS and usual care.

• No studies reported assessment of objective PA using, for example accelerometers. Validated self-report questionnaires were predominantly used.

• There was no consistent evidence of a difference at follow-up between ERS and comparator groups in respect of other outcomes, i.e. physical fitness, blood pressure, serum lipids, glycaemic control, obesity indices, respiratory function, psychological well-being and HRQoL. Only one study assessed adverse events, reporting a small increase in the rate of falls among those in both the ERS and walking programme compared with usual care.

• Although some studies reported within-group improvements (compared with baseline) in primary and secondary outcomes with ERS, these differences need to be interpreted with caution as they are subject to regression to the mean and/or a placebo/Hawthorne effect (therefore not tabularised/reported in Results section).

• None of the studies reported outcomes of ERS by disease-specific subpopulations.

Introduction

A systematic review of the literature was conducted to identify economic evidence on ERS as defined in the earlier stages of this report, i.e. schemes that involved referral from a primary health-care professional due to an underlying condition and access to a structured programme of exercise. Both economic evaluations and existing systematic reviews of economic evidence on exercise referral were considered for inclusion. By adhering to a relatively narrow definition of what constitutes ERS, a number of studies exploring the cost-effectiveness of PA were excluded on the basis that (1) they did not include a referral from a health-care professional; (2) they did not consider a population with an underlying health condition; or (3) they did not comprise a structured programme of exercise. In this respect, the findings of this economic review are intended to mirror those of the effectiveness review presented in Chapter 3 of this report.

Methods

This review was conducted and reported in accordance with the PRISMA statement. 36

Results

Identification and selection of studies

The bibliographic searches identified three economic evaluations50,61,70 of ERS that met our inclusion criteria (UK, n = 2; non-UK, n = 1). In addition, we included a model-based economic evaluation of brief interventions designed to promote PA developed to inform public health guidance issued by NICE. 76 This NICE evaluation considered ERS as one method of promoting PA in primary care. Although not published in a peer-reviewed journal, the full report of the study was available in the public domain (available at www.matrixknowledge.com/.%20.%20./physical_activity_economic_modelling_report_april2006.pdf).

In addition to the primary economic evaluations, three systematic reviews of ERS were identified,40,41,77 which included consideration of cost-effectiveness. Findings from the reviews and primary studies are reported separately below. See Figure 13 for details.

FIGURE 13.

Study inclusion process for ERS cost-effectiveness systematic review. PHC, primary health care.

Summary

• Given the lack of standardisation of the ERS definition used by previous systematic reviews and the publication of further recent evidence, we undertook a de novo systematic review of cost-effectiveness of ERS.

• Our systematic review identified only four primary economic evaluations that assessed the cost-effectiveness of ERS – three trial-based economic evaluations and a model-based analysis (commissioned by NICE as part of the development of guidance on brief interventions in primary care for the promotion of PA).

• Broadly, the previous evidence base suggests that ERS is a cost-effective intervention in sedentary, but otherwise healthy populations. However, there is some significant uncertainty around the estimates of cost-effectiveness because of an absence of evidence on the long-term effectiveness of these interventions. Although modelling studies can go some way to exploring this, ultimately these issues can only be resolved through better evidence of effectiveness derived from RCTs or other well-designed observational studies. As such, any criticism of the economic evidence should be considered in light of the evidence on effectiveness available at the time of the analysis.

• Each of the previous economic evaluations has its merits and makes a valuable contribution to the limited evidence base on the cost-effectiveness of ERS. The trial-based studies benefit from a high degree of internal consistency, deriving their estimates of effectiveness from the trial and, in some cases, detailed estimates of the cost of the interventions. Any weaknesses inherent in these analyses are also largely as a result of the limitations of the trials, particularly the degree to which the findings can be considered to be externally valid and the relatively short follow-up that was achievable in a trial setting.

• The NICE economic modelling study overcomes the issue of the short-time horizon inherent in the trial-based analyses. This study allowed for an estimate of the longer-term costs and benefits of PA, taking into account the effects on a number of long-term conditions that are known to be associated with physical inactivity. There are many weaknesses associated with the model although many of these result from an absence of evidence on the effectiveness of ERS (e.g. on the relationship between physical inactivity and long-term conditions, long-term effectiveness of interventions, adherence to interventions etc.). It should also be remembered that any economic model can only ever be a simplification of reality. In an area as complex as PA and behaviour change, and an area characterised by limitations in the evidence base, the need for simplification may be great, leading to a model that fails to meet many of the best-practice criteria.

• A further limitation of previous economic evaluations is their focus on a sedentary, but otherwise healthy population. Few of the studies explicitly consider whether or not ERS can contribute to improved outcomes in populations with underlying conditions (with the exception of Gusi et al. ,70 which was conducted outside the UK).

• In light of these findings, we decided to develop a de novo economic model to assess the cost-effectiveness of ERS. Our model builds on the principles of the NICE decision-analytic model, which includes some important further development of the methods and a more robust approach to the incorporation of ERS effectiveness evidence. The findings of this analysis are presented in Chapter 6.

Background

Chapter 3 provides an overview of the effectiveness of ERS. However, if patients do not initially take up an exercise referral then the beneficial effects of increased PA will not occur, or, conversely, greater adherence to ERS will increase the probability of being physically active. Public health impact depends on ‘real-world’ effectiveness information and is therefore dependent not only on RCT evidence, but also on the external validity of this evidence. With widening health inequalities, those most at need may also be those most likely to have lower uptake and adherence to ERS. Further, the costs related to ERS are ‘front loaded’, so where patients fail to attend or drop out, this will reduce the cost-effectiveness of ERS. Therefore, it is important to understand the patient-level factors and programme-level factors that might influence uptake and adherence to ERS.

The objectives of this systematic review are to (1) quantify the levels of uptake and adherence to ERS; (2) identify demographic and medical diagnosis variables, programme factors and psychosocial factors (e.g. self-determination) that predict uptake and adherence to ERS; and (3) identify from qualitative studies patient perceptions about recruitment, referral and ERS engagement processes, and associated benefits.

Variation in individual ERS programmes and the variable reporting and monitoring of patients characteristics related to uptake and adherence85 may explain the lack of standard definitions for ERS uptake and adherence in the literature. For the purposes of this systematic review, the following definitions were used:

• Uptake The proportion of those individuals offered entry to ERS who attend an initial consultation with an ‘exercise professional’ or attend a first exercise session.

• Adherence Of those individuals who take up ERS, what proportion experience at least 75% of the programme.

Methods

This review was conducted and reported in accordance with the PRISMA statement. 36

Results

Identification and selection of studies

Of the 233 full papers retrieved from the Chapter 3 search and identification through other means (reference list check, author and expert knowledge), five RCTs, 14 observational studies and 10 qualitative studies (28 primary studies in total) and two systematic reviews were judged to meet the inclusion criteria. Figure 14 summarises the selection process.

FIGURE 14.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow diagram of study inclusion process for predictors of uptake and adherence to ERS systematic review. PHC, primary health care.

Findings of quantitative primary studies

Sample sizes ranged across studies from 30 to 6610 participants in the observational studies, and from 97 to 363 participants in the RCTs. Mean age ranged from 44.9 to 51.9 years across the observational studies and from 53.9 to 59.1 years for RCTs. Across the 19 studies,27,28,50,61,69,88,91,93,95,97–100,102–107 12 provided a definition of uptake (Table 33). 88,89,93,95,97–100,102–107 Uptake was defined in one of two ways: attendance at the initial consultation with the ‘exercise professional’ or attendance at at least one exercise session. Thirteen studies26,60,87,90,91,96,97,99–106 provided a definition of adherence – completion of a set number of exercise sessions, either numerically (e.g. completed 20 sessions92) or as a percentage (e.g. > 80% attendance104). For four studies,88,91,92,103 attendance at a post-ERS consultation was also required to meet the definition of adherence.

TABLE 33 - Characteristics of studies reporting levels and/or prediction of uptake and adherence to ERS

F, finishers; NF, non-finishers; IHD, ischaemic heart disease; ref., reference.

Two schemes evaluated: schemes A and B.

Three publications from one study; for ease of reading ref. 103 will subsequently be adopted.

Study	Population characteristics: mean age (years), gender (% male), medical diagnoses or risk factors (%)	Inclusion/exclusion criteria of study	Sample size	ERS setting	Uptake definition	Adherence definition
RCTs
Taylor et al. (1998)27 UK	Mean age: 54.1 Male: 37 Smokers: 43 Overweight: 77 Hypertensive: 46	Smokers, hypertension (at least 140/90 mmHg), overweight (BMI > 25)	97	Three practices	Attended at least one session	Patients who attended at least 15 sessions
Stevens et al. (1998)50 UK	Mean age: 59.1 Male: 40 Smoker: 18	Sedentary: < 20 × 30 minutes of moderate-intensity PA or less than 12 × 20 vigorous-intensity PA in the last 4 weeks	363	One practice	‘Attended initial consultation with exercise development officer’	Not reported
Harrison et al. (2005)28 UK	Mean age: not reported Male: 33 Smoker: 24.4 At least one CHD risk factor: 75.3	Scheme related	275	46 practices	‘Attended the first exercise consultation’	Not reported
Isaacs et al. (2007)61 UK	Mean age: 57.1 Male: 35 Raised cholesterol: 24.0 Hypertension: 44.5 Obesity: 65.9 Smoking: 10.4 Type 2 diabetes: 12.3/11.3 Family history of MI: 13.9	Not active (no definition reported), raised cholesterol, controlled mild/moderate hypertension, obesity, smoking, diabetes, family history of MI at early age	317	88 practices	Attended at least one session	The adherence of subjects to the active intervention arms was assessed by the use of handheld diaries and class registers (75–100% adherence)
Sorensen et al. (2008)69 Denmark	Mean age: 53.9 Male: 43 Metabolic syndrome: 36 Type 2 diabetes: 18 CVD: 32 Other diseases: 14	Scheme related	28	14 practices	Correspondence with author	Not reported
Observational studies
Damush et al. (2001)93 (prospective) USA	Mean age: 64.1 (9.1) Male: 0 COPD: 13.1 Congestive heart failure: 14.9 Coronary artery disease: 17.5 Hypertension: 90.5 Type 2 diabetes: 38.9 History of stroke: 13.1	Female, age ≥ 50 years, not terminally ill, visited health centre in previous 12 months and had a scheduled or walk-in visit during the 6-month enrolment period	404	Two health centres	‘Participation in at least one exercise class’	Not reported
Dinan et al. (2006)97 (prospective) UK	Aged 75 years and over	Scheme related	242	14 practices	‘Took up the referral’	‘Completed the cycle of exercise classes’
aDugdill et al. (2005)95 (prospective) UK	Mean age: not reported A – male: 36 B – male: 41 A – overweight: 37 A – hypertension: 13 A – mental illness: 9 B – arthritis: 28 B – back pain: 26 B – overweight: 23	Scheme related	A: 980 B: 1825	Two schemes	Attended first consultation with exercise officer	Not reported
Edmunds et al. (2007)98 (prospective) UK	Mean age: 44.98 (14.61) Male: 16 Medical diagnoses: not reported	Overweight or obese	49	One scheme	Not reported	1–5 scale, an individual was defined as having ‘dropped out’ if he/she had stopped participating in their prescribed activities at their exercise referral site/facility
Harrison et al. (2005)99 (prospective) UK	Mean age: 51.3 (12.6) Male: 39.2 Musculoskeletal problems: 32.8 CVD: 29.9 Overweight: 10.4 Fitness: 5.8 Mental-health problems: 5.1 Respiratory: 4.1 Other: 0.7	Scheme related	6610	One scheme	Attended first consultation with exercise officer	Not reported
Jackson et al. (1998)91 (retrospective) UK	Mean age: not reported Gender: not reported Medical diagnoses: not reported	Not reported	686	One scheme	Not reported	Adherers: exercised at the leisure centre over a 10-week period and attended a 10-week consultation Non-adherers: discontinued exercise at the leisure centre within the 10-week period and did not attend a 10-week consultation
Jones et al. (2005)100 (prospective) UK	Mean age: not reported Male: 42.1 Medical diagnoses: not reported	High blood pressure, weight or stress-related problems (or combinations of these)	152	One scheme, seven leisure centres	‘Attended their local gym for an initial assessment’	Those who completed 24 sessions in total were considered to be successful
Lord and Green (1995)88 (prospective) UK	Mean age: not reported Male: 25 Overweight: 32.2 Stress/anxiety: 15 Other: 11 Lipids/cholesterol: 6.4 Keep fit: 4.8 Lack of exercise: 4.8 Depression: 4.8 Arthritis: 2.9 Back pain: 2.9 Family history of IHD: 2.4	Scheme related	419	One scheme	‘Attended an initial consultation with the community health and fitness officer’	Compliance: ‘those participants who returned to attend a 10-week consultation and who were still exercising’
Martin and Woolf-May (1999)92 (retrospective) UK	Mean age (F): 51.1 (12.3) Mean age (NF): 54.7 (14.4) Male: 35 BMI > 25 (F): 50 Cholesterol > 5.2 mmol/l (F): 16.7 Family history of CVD/CHD (F): 23.8 BMI > 25 (NF): 51.4 Cholesterol > 5.2 mmol/l (NF): 2.9 Family history of CVD/CHD (NF): 17.1	Scheme related, 60 NFs randomly selected to match F’s number	490	One leisure centre	Not reported	Finishers: completed 20 sessions over the 10-week period and/or completed the final assessment Non-finishers: completed less than 20 sessions over the 10-week period
Morton et al. (2008)101 (prospective) UK	Mean age: 51.9 (15.7) Male: 26.7 Medical diagnoses: not reported	Not reported	30	One leisure centre	Not reported	Adherers: ‘attended at least one session per week for the duration of the study’ Partial adherers: ‘attended intermittent sessions, but specifically stated that they had not dropped out of the scheme’ Dropped out: ‘attended’ no sessions or made personal contact with the leisure centre to terminate their involvement with the scheme
Roessler and Ibsen (2009)102 (prospective) Denmark	Mean age: not reported Male 33 BMI > 25: 35 BMI > 30: 37 BMI > 35: 28	Scheme related	1156	One scheme	Not reported	Completed intervention (not defined)
Sowden et al. (2008)103 (retrospective) UK	Mean age: 51 Male: 35 CVD risk: 44.3 Overweight/obese: 30 Musculoskleletal:25.2 Mental health: 19.8 Diabetes: 17.7 Respiratory: 8.1	Scheme related	6101	Six schemes, 317 practices	‘Attendance at initial appointment’	‘Attends final ERS appointment’
James et al. (2009)104 (prospective) UK	Mean age: not reported Male: 34.6 Metabolic diseases: 36.3 Orthopaedic diseases: 24.7 CHD: 17.5 Pulmonary diseases: 9.9 Mental health: 9.4 Neuromuscular diseases: 1.1 Others: 1.1	Scheme related	1315	One scheme, five leisure centres	Not reported	Completers: ‘attended > 80% of scheduled sessions’
bGidlow et al. (2007),105 Crone et al. (2008),106 James et al. (2008)107 (prospective) UK	Mean age: 50.8 (14.4) Male: 39.9 Obesity: 30.3 Musculoskleletal:26.3 CVD: 16 Mental health: 4.6	Scheme related	2864–2958	One scheme	‘Attendance of at least one session’	Completers: ‘attended > 80% of scheduled sessions’

Exercise referral schemes uptake and adherence levels

The pooled level of ERS uptake across the observational studies was 66.% (95% CI 57% to 75%) compared with 80% (95% CI 61% to 98%) across the RCT (Figure 15). There was a high level of statistical heterogeneity for both observational studies (I² = 99.4%, p < 0.0001) and RCTs (I² = 99.3%, p < 0.0001). The studies of Stevens et al. 50 and Damush et al. 93 reported particularly low levels of uptake, i.e. < 35%. Stevens et al. 50 hypothesise that the low uptake they experienced may have been reflective of the nature of the invitation letter sent to participants and the point of randomisation (pre-invitation letter). They also hypothesise that a change in the format of the letter (e.g. including a specific date offered for the first ERS appointment) would have improved participation. Similarly, Damush et al. 93 used a letter-based recruitment because studies conducted in the USA require a pre-exercise test before the exercise intervention commences, and potentially this could have deterred eligible patients from consenting to the study.

FIGURE 15.

Meta-analysis of ERS uptake levels stratified by study design. Overall test for heterogeneity between subgroups: 691.58 [degrees of freedom (df) 1], p = 0.28. Note: weights are from random-effects analysis.

Levels of adherence to ERS were variable across all study types (range 12–93%) (Table 34). The pooled level of ERS adherence was 49% (95% CI 40% to 59%) for observational studies and 37% (95% CI 20% to 54%) for the RCTs (Figure 16). Again, there was a high level of statistical heterogeneity for both observational studies (I² = 99.1%, p < 0.0001) and RCTs (I² = 89.0%, p <0.003). The observational study by Martin and Woolf-May92 reported particularly low levels of adherence (12%). Unfortunately, the study publication does not provide sufficient information on the ERS process to allow an appraisal of its contribution to the low adherence rate. The authors stated that ‘of the available 490 subjects there were only 60 known finishers’, suggesting that some individuals who may have adhered may have been missed.

TABLE 34 - Summary of uptake and adherence to ERS levels across studies

Two schemes evaluated: schemes A and B.

Average of the three publications.

Study	Uptake	Adherence
	% (n/N)	% (n/N) of patients who took up ERS	% (n/N) of patients who were referred to ERS
RCTs
Taylor et al. 27 UK	88% (85/97)	28% (24/85)	25% (24/97)
Stevens et al. 50 UK	35% (126/363)	Not reported	Not reported
Harrison et al. 28 UK	84% (232/275)	Not reported	Not reported
Isaacs et al. 61 UK	92% (293/317)	45% (133/293)	42% (133/317)
Sorensen et al. 69 UK	100% (28/28)	Not reported	Not reported
Observational studies
Damush et al. 93 USA	28% (113/404)	Not reported	Not reported
Dinan et al. 97 UK	89% (216/242)	82% (178/216)	74% (178/242)
aDugdill et al. 95 UK	B: 68% (1825/2696)	A: 34% (336/958) B: 46% (849/1829)	B: 32% (849/2698)
Edmunds et al. 98 UK	Not reported	51% (25/49)	Not reported
Harrison et al. 99 UK	79% (5225/6610)	Not reported	Not reported
Jackson et al. 91 UK	Not reported	70% (466/686)	Not reported
Jones et al. 100 UK	78% (119/152)	65% (77/119)	51% (77/152)
Lord and Green88 UK	60% (252/419)	31% (77/252)	18% (77/419)
Martin and Woolf-May92 UK	Not reported	12% (60/490)	Not reported
Morton et al. 101 UK	Not reported	40% (12/30)	Not reported
Roessler and Ibsen102 Denmark	Not reported	70% (811/1156)	Not reported
Sowden et al. 103 UK	58% (3565/6101)	39% (1404/3565)	23% (1404/6101)
James et al. 104 UK	Not reported	57% (750/1315)	Not reported
bGidlow et al. ,105 Crone et al. ,106 James et al. 107 UK	66% (1930/2908)	48% (931/1930)	32% (931/2908)

FIGURE 16.

Meta-analysis of ERS adherence levels stratified by study design. Overall test for heterogeneity between subgroups: 4.57 (df 1), p = 0.397. Note: weights are from random-effects analysis.

Predictors of participant uptake

Six observational88,93,95,99,103,105 and two RCTs27,61 reported predictors of uptake, with five studies providing bivariate analysis27,61,88,95,105 and four studies multivariate analysis93,99,103,105 in which associations among predictor variables were adjusted for other factors such as age.

Demographic

The findings of studies that assessed demographic predictors of ERS uptake are summarised in Table 35. Two studies88,103 reported that females were more likely to take up ERS than men, whereas two studies99,105 showed no association of gender with uptake. Increasing age was positively associated with increased levels of ERS uptake in three studies,95,103,105 whereas three studies61,88,93 found no such association. Gidlow et al. 105 found that those who were the most deprived were less likely to uptake. Gidlow et al. 105 also found that those individuals living in a more rural location were less likely to take up ERS. Damush et al. 93 found no association between ethnicity and uptake.

TABLE 35 - Summary of analysis of predictors of ERS uptake

0, no statistically significant association between variable and level of uptake; +, a positive statistically significant (p < 0.05) association between variables and level of adherence, e.g. increased age of participants was associated with increased levels of uptake; –, a negative statistically significant (p < 0.05) association between variables and level of adherence, e.g. increased age of participants was associated with decreased levels of uptake.

Six geographical areas of London, UK.

Two neighbourhood community health centres located in Indianapolis, IN, USA.

Study	Multivariate analysis				Bivariate analysis			Overall summary
Variables	Harrison et al.99 UK	Sowden et al.103 UK	Gidlow et al.105 UK	Damush et al.93 USA	Lord and Green 88 UK	Dugdill et al.95 UK	Isaacs et al.61 UK	No. of studies on this factor	Any significant association		No significant association
Sample size	n = 6610	n = 6101	n = 2958	n = 404	n = 419	n = 2696	n = 317
									+	–
Demographic factors
Gender (male vs female)	0	+	0		+			4	2	0	2
Increasing age		+	+	0	0	+	0	6	3	0	3
Deprivation	0	0	–					3	0	1	2
Rurality (urban vs rural)			–					1	0	1	0
Ethnicity (African American vs all other racial groups)				0				1	0	0	1
Programme factors
Scheme locationa		0						1	0	0	1
Clinic locationb				0
Referrer (GP vs ‘other’)			–					1	0	1	0

Medical diagnosis

Given the variable way in which referral reason (medical history) was analysed and reported, it was not possible to tabularise this in a summary way. Harrison et al. 99 showed found that those with mental health problems (OR 1.79, 95% CI 1.24 to 2.39, p < 0.01) or fitness needs (OR 10.33, 95% CI 1.44 to 74.3, p < 0.05) were more likely to take up ERS than those with no specified referral reason. Harrison et al. 99 also reported that those with respiratory problems and most deprived were more likely to take up ERS than those with respiratory problems and least deprived (OR 1.45, 95% CI 1.06 to 1.99, p < 0.05). Gidlow et al. 105 showed those patients referred with mental-health (OR 0.33, 95% CI 0.27 to 0.57, p < 0.01), musculoskeletal (OR 0.75 95% CI 0.58 to 0.99, p < 0.05), overweight/obesity (OR 0.63, 95% CI 0.50 to 0.81, p < 0.01) or ‘other’ (not defined) (OR 0.63, 95% CI 0.46 to 0.85, p < 0.01) problems were less likely to take up ERS than patients with cardiovascular disease. In contrast, Sowden et al. 103 found that those referred with a musculoskeletal (OR 1.18, 95% CI 1.01 to 1.38, p < 0.05) problem were more likely to take up ERS. This was not the case for those with diabetes or CVD. Gidlow et al. 105 reported that more patients referred for mental-health problems took up ERS than those referred for physical-health problems (60% vs 69%; p < 0.001). Taylor27 found that more individuals referred for obesity took up ERS than those referred for smoking (p < 0.01).

Programme factors

Gidlow et al. 105 reported that GP referrals were more likely to lead to uptake than referral by another individual. Sowden et al. 103 and Damush et al. 93 both observed no association between for scheme and clinic location and uptake.

Predictors of exercise referral scheme adherence

Eight observational88,95,98,100,101,103–105 and two RCTs27,61 reported predictors of adherence. Seven studies undertook bivariate statistical analysis27,61,88,95,100,101,105 and four studies undertook multivariate statistical analysis. 98,103–105

Demographic

The findings of studies that assessed demographic predictors of ERS adherence are summarised in Table 36. Two studies95,105 reported that men are more likely to adhere than women, while three studies88,103,104 found no such association. Increasing age was a predictor of increased ERS adherence in five studies,88,95,103–105 although two studies showed no association with age. 61,88 Deprivation, rurality, referrer, leisure provider105 and occupation104 were all found not to be significant predictors of ERS adherence. Dugdill et al. 95 found that fewer patients adhered to ERS (p < 0.01) when referred by the GP (32%) compared with a practice nurse (45%) or a cardiac nurse (57%). James et al. 104 reported that those of mixed ethnicity were more likely to adhere to an ERS.

TABLE 36 - Summary of analysis of predictive factors of ERS adherence

+, a positive statistically significant (p < 0.05) association between variables and level of adherence, e.g. increased age of participants was associated with increased levels of adherence; –, a negative statistically significant (p < 0.05) association between variables and level of adherence, e.g. increased age of participants was associated with decreased levels of adherence; 0, no statistically significant association between variable and level of adherence.

Increasing age from 18 to 54 years only, not for ≥ 55 years.

Study	Multivariate analysis				Bivariate analysis					Overall summary
		Gidlow et al.105 UK
Variables	Sowden et al.103 UK	James et al.107 UK	James et al.104 UK	Edmunds et al.98 UK	Lord and Green88 UK	Dugdill et al.95 UK	Isaacs et al.61 UK	Jones et al.100 UK	Morton et al.101 UK	No. of studies on this factor	Any significant association		No significant association
Sample size	n = 3565	n = 1996	n = 1315	n = 49	n = 419	n = 2696	n = 317	n = 119	n = 30
											+	–
Demographic factors
Gender (male vs female)	0	–	0		0	–				5	0	2	2
Increasing age	+	+	+		0/+a	+	0			6	4/5	0	1/2
Deprivation	0	0	0							3	0	0	3
Rurality (urban vs rural)		0								1	0	0	1
Occupation			0							1	0	0	3
Leisure provider		0								1	0	0	1
Referrer (GP vs other health-care professionals)		0				+				2	0	1	1
Ethnicity (white vs mixed)			+							1	1	0	0
Psychosocial factors
Stage of change								0		1	0	0	1
Self-efficacy								0		1	0	0	1
Expectations of change (health and fitness)								0		1	0	0	1
Expectations of change (personal development)								+		1	1	0	0
Psychological well-being								0		1	0	0	1
Need satisfaction				0						1	0	0	1
Perceived autonomy				0						1	0	0	1
Support
Self-determination				0					+	2	1	0	1

Programme factors

Sowden et al. 103 reported variable levels of patient adherence across six different ERS areas in the London area. This finding illustrates the potential influence of programme-level factors on adherence.

Medical diagnosis

In the James et al. study,104 patients with pulmonary problems were less likely to adhere than those with CVD. Sowden et al. 103 reported that patients with diabetes were less likely to adhere to an ERS (OR 0.76, 95% CI 0.63 to 0.93, p < 0.01) than those with CVD (OR 1.22, 95% CI 1.03 to 1.45, p < 0.05) when both compared with those without either condition.

Gidlow et al. 105 found that those referred for mental health problems were less likely to adhere to ERS than those referred for physical health problems (22% vs 34%, p < 0.001). Taylor et al. 27 reported no difference in adherence between those individuals who were referred because they were a smoker, overweight, obese or hypertensive.

Psychosocial

Three studies98,100,101 assessed the psychosocial predictors of adherence (Table 36). Morton et al. 101 found participant self-determination to positively predict ERS adherence, whereas Edmunds et al. 98 found no such association. An expectation for change in personal development was also found to be positively predictive of ERS adherence. 100

Programme factors

No programme factors were reported in studies examining associations of ERS adherence.

Summary

There has been little consideration of uptake and adherence in previous systematic reviews of ERS.

• Fourteen observational studies and five RCTs reported their level of ERS uptake (the proportion of those individuals offered entry to ERS who attend an initial consultation with an ‘exercise professional’ or attend a first exercise session) and/or adherence (of those that uptake ERS, what proportion undertake 75–100% of the programme) (UK, n = 16; non-UK, n = 3).

• The pooled estimate for ERS uptake across the observational studies (66%) appeared to be lower than the pooled estimate for RCTs (80%). The pooled estimate for ERS adherence in the observational studies (50%) appeared to be higher than the pooled estimate for RCTs (37%). However, it is important to note that there was a high degree of statistical heterogeneity in the levels of uptake and adherence across studies.

• Only 6 of 13 included studies undertook multivariate analysis to assess the association between potential predictors and levels of uptake or adherence, i.e. adjusted for potential confounders. The remaining seven studies undertook bivariate association analysis.

• Although a number of studies reported an association between participant gender or age and ERS uptake and adherence, very few studies reported associations for psychosocial and programme-level factors, for example the time of day ERS is available at the delivery site.

• Women and older people were more likely to take up ERS. Although older people were also more likely to adhere, women were less likely to adhere than men.

• Eleven qualitative studies highlighted the complexity of personal experiences with ERS that might influence uptake and adherence. Several critical factors reflected the importance of individualised support that takes account of low levels of confidence. However, logistic factors such as cost, convenience and child support were also important to some population sectors.

Introduction

There is limited evidence on the cost-effectiveness of ERS. The available evidence highlights significant uncertainty, particularly around the effectiveness of ERS. The result is that decision-makers are currently making decisions on the availability of ERS with only limited evidence on its cost-effectiveness.

In light of this, a de novo analysis has been developed to further explore the cost-effectiveness of ERS. The analysis considers a target population of sedentary adults, with further analysis presented to explore the impact of ERS on those with specific pre-existing conditions, where evidence suggests that ERS might improve outcomes. The approach taken uses previous research as a point of departure, and builds on this through use of evidence synthesis (see Chapter 3) and through further analysis of the impact of PA on HRQoL.

The approach here comprises three main activities:

1. The development of a cost–utility analysis, similar to earlier analyses, to estimate the impact of ERS on long-term outcomes based on the effectiveness evidence identified herein, including subgroup analysis, to explore the cost-effectiveness of ERS in individuals with pre-existing conditions.

2. The development of methods to quantify and incorporate short-term benefits of PA into this cost–utility framework.

3. A cost–consequence framework that summarises the costs and benefits associated with ERS in a disaggregated fashion.

Methods for cost–utility modelling approach

Modelling approach

Figure 17 illustrates our modelling approach, which is a based on the structure of the model developed by NICE. 76 A decision-analytic model was developed, which followed a cohort of individuals over time to examine the impact of PA on their health. Specifically, the model considered the lifetime risk of developing a series of conditions that are known to be associated with being physically active. The model considered the impact of ERS on coronary heart disease, stroke and type 2 diabetes, because these are considered to be the conditions for which the most robust quantifiable evidence is available on the relationship between PA and incidence of disease. Furthermore, evidence on the QALY losses associated with the development of these conditions is also available from previous research. 84 PA has been associated with a wide range of conditions. Owing to data limitations, no attempt was made to incorporate the effect of PA on other conditions, such as musculoskeletal or respiratory diseases.

FIGURE 17.

Model structure.

The model considers a cohort of individuals, aged between 40 and 60 years, who present in a sedentary state. The age of the population was selected to reflect the evidence on the clinical effectiveness of ERS reported in Chapter 3. Individuals enter the model as either exposed to an ERS intervention or not; modelling considers two hypothetical cohorts, comparing costs and outcomes of a cohort exposed to ERS with a control cohort not exposed to ERS. Those exposed to ERS are assumed to have a greater probability of becoming active. A physically active individual is assumed to have both improved life expectancy and quality of life (QoL), as a result of a reduced risk of developing each of the morbidities considered in the model. The primary end point for the analysis was QALYs.

The intervention

The ERS intervention in the model is consistent with the definition used throughout this report (see Chapter 1, Physical activity promotion in primary care). Effectiveness data for ERS are derived from the meta-analysis presented herein (Figure 3). For the purposes of our analysis, we assume that the ERS is leisure centre based, as is the case for the majority of studies considered in Chapter 3. Estimates of the cost of the intervention are derived on this basis.

Comparator

The comparator for the analysis is ‘usual care’, which is specified as no active intervention and as the recognised alternative in a sedentary population. This acknowledges that some sedentary individuals may choose to participate in PA without an intervention, although the probability of doing so is assumed to increase as a result of exposure to an intervention.

Perspective

The model adopts a NHS/Personal Social Services perspective, in line with the NICE reference case for cost-effectiveness analysis. 121 Although it is acknowledged that PA may have important effects on non-health-care costs and benefits, these are excluded from the primary/base-case cost–utility analysis, although these broader considerations are addressed in sensitivity analysis and through the presentation of cost–consequence analyses.

Time horizon

A lifetime horizon is adopted to acknowledge the long-term benefits of PA, with alternative time horizons considered in sensitivity analysis.

Model inputs

Data on costs and effects were synthesised to populate the model. Data were primarily derived from the systematic reviews undertaken in Chapters 3 and 4. Further details are provided below.

Effectiveness of exercise referral scheme comparator

Evidence of the effectiveness of ERS/comparator, measured in terms of the probability of moving from a sedentary state to an active state, was derived from the meta-analysis conducted as part of clinical effectiveness review in Chapter 3. This was based on ITT analyses, which adjusted for adherence and uptake and showed ERS to be associated with a higher probability (RR 1.11, 95% CI 0.99 to 1.25) of being active compared with usual care (Figure 3). The active state is defined in line with the effectiveness literature, i.e. doing 90–150 minutes of at least moderate-intensity PA per week. Thus, a sedentary lifestyle corresponds not only to non-participation in PA but also to participation below the requisite amount. The active state is assumed to last long enough to enable health benefits to be obtained, although this remains undefined given the inadequate evidence on the dose–response relationship between PA and the incidence of long-term outcomes. Previous analyses of behaviour change have referred to this scenario as ‘fully engaged’122 to describe an individual who makes lasting changes to his or her lifestyle following an intervention.

Risks of developing health states associated with inactivity

Evidence of the effect of PA on the development of the outcomes considered in the model (CHD, stroke and type 2 diabetes) is derived from a systematic review of economic evaluations in Chapter 4 and HSE – 2006. 123 The derivation of the estimates involved a number of steps. First, the probability of developing these conditions among sedentary individuals was generated from the prevalence of these conditions in that population using the HSE – 2006123 data. Although it is acknowledged that a potential limitation of such univariate analyses is that it does not adjust for confounders, data constraints precluded the inclusion of those confounders. The second step involved estimating the probability of developing the health states among active individuals using RR estimates identified from NICE76 to adjust the estimates derived from the first step. It must be emphasised that the PA levels and study population used to measure the RR estimates match those identified in our clinical effectiveness review. A number of assumptions were made in generating these estimates. First, the risk estimates were assumed to be equivalent to the risk of developing those conditions over a lifetime. Second, the risk of experiencing any of these health states was assumed to be independent of the risk of experiencing other health states. Third, individuals were assumed to experience only one health state within the model.

Exercise referral scheme intervention costs

The cost of the ERS intervention was derived from previously published research identified as part of the review conducted for this study. The study by Isaacs et al. ,61 presenting a detailed bottom-up costing exercise, was identified via a systematic review of the literature, and is regarded here as the best available evidence/estimate for costing of ERS. The estimated cost of the intervention was based on resource use in a health service and/or local authority setting, consistent with the primary perspective taken for analyses here. See Table 26 for further details (information of the calculation of these costs can be found in Isaacs et al. 61 The validity of the costs estimates was assessed by the expert advisory group on this project and judged to be representative of ERS schemes currently in operation. The cost estimates were adjusted for inflation into 2010 prices using the Consumer Price Index. Discounting of the intervention costs was not undertaken as intervention costs were assumed to be wholly incurred in the first year. No attempt was made to estimate a net cost of the intervention, which subtracts any cost savings that might result from ERS from the cost of the intervention. Where this was explored in the systematic review in Chapter 4 (Isaacs et al. ,61 Gusi et al. 70), there was no clear evidence of a change in health-care utilisation (e.g. medications, hospital or primary care) as a result of the intervention.

Treatment costs and quality-adjusted life-years associated with coronary heart disease, stroke and type 2 diabetes

The model considers three outcomes associated with PA, CHD, stroke and type 2 diabetes. The total lifetime treatment costs and QALYs associated with each condition were estimated based on assumptions relating to the age at onset and the likely life expectancy combined with estimates of the annual cost of treating an individual with the condition. This approach was in line with the earlier analysis conducted by NICE. 76

It was assumed that the treatment cost of stroke, unlike the other health states was an event cost that occurs once, rather than a recurring cost. This is acknowledged as a simplification in the model, as in reality there are likely to be acute and ongoing costs associated with stroke. Treatment costs were discounted using the prevailing discount rates as determined by the Treasury and/or NICE guidelines (i.e. 3.5% discounting rate).

Primary outcome measure (quality-adjusted life-years)

The primary outcome of the economic evaluation is expressed in terms of QALYs. QALY losses associated with each of the conditions considered in the model are calculated. QALYs were discounted at 3.5% discount rate. The formula for calculating the QALYs is:

^{[Equation 1]} Q = Q 1 ( t s ) + Q 2 ( t 3 – t 4 )

where Q₁ = mean QoL associated with being in a non-disease health state; Q₂ = mean QoL associated with a particular disease health state; t_s = number of years before onset of the disease health state (average age minus 55 years); t₃ = age at disease health state onset and t₄ = mean age of mortality associated with health state (average age of mortality minus loss of life-years associated with the particular condition). Loss of life-years was calculated by subtracting life-years remaining after onset of the disease health state from the average life-years remaining for the non-disease health state.

Assessment of uncertainty

Uncertainty in parameter estimates was explored through the use of deterministic and probability sensitivity analyses. The deterministic sensitivity analysis, which covered one-way and scenario analysis, explored a number of uncertainties that were recognised at the outset of the analysis. These included uncertainties around the effectiveness of ERS and changes in the cost of ERS to take into account costs incurred by participants as well as providers. The effectiveness of ERS was varied according to estimates of uncertainty reflected in the upper and lower limits of the 95% CI of the RR estimate. Sensitivity analysis also considered how a less intensive form of ERS might look, using evidence on a walking-based intervention (as opposed to a structured leisure centre-based intervention) from Isaacs et al. 61 Further sensitivity analyses considered ‘best-case’ and ‘worst-case’ scenarios that considered the combined effect of extreme values of effectiveness and cost.

In addition, uncertainties around parameters considered to be key drivers of the cost-effectiveness of ERS were addressed simultaneously using PSAs. The parameters that had different unit values in the two arms of the model (i.e. probability to be active and probability to get the disease conditions) were specified as incremental differences between the two arms and not absolute values. The intuition is that the distributions of these parameters may be correlated and, hence, representing them as absolute values may overestimate the uncertainty. The distributions and the calculation of alpha and beta calculations were based on Briggs et al. 124 In cases where there were no data on standard errors (SEs), the standard approach of using 10% of mean estimates as SE was followed. A total of 10,000 Monte Carlo simulations were generated from the PSA.

Model validation

The following procedures were employed to check the validity of the model (Chilcott et al. 125):

1. Internal validation Simulate a series of changes in the input values that are likely to vary the results of the model with checks to see that the impacts on the results are expected. For example, setting all QALY parameters to zero, and checking if the output of the QALYs in each arm is zero. In addition to this, the model was reviewed by an experienced health economist who was not part of the research team.

2. Peer review A peer-review process that involved a modeller, who understands the complexities of the model, scrutinising the spreadsheet of the model and the formulae behind it.

Results

Deterministic sensitivity analysis

Deterministic sensitivity analysis was carried out around parameters with known uncertainty. Sensitivity analyses conducted are summarised in Table 44. Table 45 shows the impact of the variation in parameter estimates (one-way analysis) on the cost-effectiveness of ERS. Assuming a less intensive ERS or more effective ERS resulted in an ICER below £30,000 and lower than the base case. On the other hand, including intervention costs to participants led to an ICER above £30,000, although a less effective ERS resulted in ERS being dominated by usual care (negative ICER) – i.e. ERS is more expensive and leads to loss of health gains.

TABLE 44 - Deterministic one-way sensitivity analysis inputs

In 2010 prices.

Parameters	Value	Data source	How data was adjusted for in the model
Intervention costs to participants	£120a	Isaacs et al. (2007)61	Costs of intervention varied from £222 to £342 (including costs to providers and participants)
Less intensive ERS	£110a	Isaacs et al. (2007)61	Costs of intervention was varied from £222 to £110
Effectiveness of ERS (based on lower limit of 95% CI)	0.294	Meta-analysis in Chapter 3	Probability of becoming active after exposure to ERS was varied from 0.336 to 0.294
Effectiveness of ERS (based on upper limit of 95% CI)	0.371	Meta-analysis in Chapter 3	Probability of becoming active after exposure to ERS was varied from 0.336 to 0.371

TABLE 45 - Cost-effectiveness results (after one-way sensitivity analyses) comparing ERS with usual care

Negative ICERs are not reported to avoid erroneous interpretation.

Parameter	Incremental cost per person (£)	Incremental effect per person (QALY)	ICER (£)
Base-case assumptions	169.54	0.008	20,876.27
Intervention costs to participants	289.54	0.008	35,652.46
Less intensive ERS	57.54	0.008	7085.16
Effectiveness of ERS (based lower limit of 95% CI)	226.04	–0.001	Dominateda
Effectiveness of ERS (based upper limit of 95% CI)	122.46	0.015	7947.11

Further analyses were conducted which considered ‘best-case’ and ‘worst-case’ scenarios for ERS. These scenarios are summarised in Table 46. The findings of the analysis are presented in Table 47. In the worst-case scenario, ERS was dominated by the comparator. In the best-case scenario, the ICER fell to under £700 per QALY. These findings of the deterministic sensitivity analysis (excluding the dominated cases) are presented in the form of a tornado diagram (Figure 18) to illustrate the relative magnitude of effect of changing each of the parameter values or scenarios. Overall, the cost-effectiveness was found to be most sensitive to changes in the scenarios (best cases of cost and effectiveness).

TABLE 46 - Deterministic scenario sensitivity analysis inputs

Scenarios	Description
Worst case	Worst-case cost (£342) and worst-case effectiveness (0.294)
Best case	Best-case cost (£110) and best-case effectiveness (0.371)
Interaction between worst and best cases	(1) Best-case cost (£110) and worst-case effectiveness (0.294) (2) Worst-case cost (£342) and best-case effectiveness (0.371)

TABLE 47 - Cost-effectiveness results (after scenario sensitivity analyses) comparing ERS with usual care

Negative ICERs are not reported to avoid erroneous interpretation.

Scenarios	Incremental cost per person (£)	Incremental effect per person (QALY)	ICER (£)
Base-case assumptions	169.54	0.008	20,876.27
Worst cases of cost and effectiveness	346.04	–0.001	Dominateda
Best cases of cost and effectiveness	10.46	0.015	678.82
Worst-case cost and best-case effectiveness	242.46	0.015	15,734.56
Best-case cost and worst-case effectiveness	114.04	–0.001	Dominateda

FIGURE 18.

Impact of deterministic sensitivity analysis on base case ICER (£20,876.27).

Probabilistic sensitivity analysis

Probabilistic sensitivity analysis, based on 10,000 simulations, was also conducted. A summary of the distributions adopted in the PSA is presented below in Table 48.

TABLE 48 - Probabilistic sensitivity analysis inputs

Parameters	Mean	SE	Distribution	Alpha	Beta
Incremental probability to be active	0.039	0.0039	Beta	96.061	2367.042
Incremental probability to experience CHD	0.013	0.0013	Beta	98.687	7492.621
Incremental probability to experience stroke	0.004	0.0004	Beta	99.596	24,799.4
Incremental probability to experience diabetes	0.022	0.0022	Beta	97.778	4346.677
Treatment discounted cost of CHD	£17,728.03	£1772.803	Gamma	100	177.2803
Treatment discounted cost of stroke	£1965.165	£196.5165	Gamma	100	19.65165
Treatment discounted cost of diabetes	£50,309.43	£5030.943	Gamma	100	503.0943
Discounted QALY for CHD health state	9.942348	0.994235	Gamma	100	0.099423
Discounted QALY for stroke health state	5.148217	0.514822	Gamma	100	0.051482
Discounted QALY for type 2 diabetes health state	14.18193	1.418193	Gamma	100	0.141819
Cost of intervention	£222	£37.9	Gamma	34.31054	6.470315

A scatterplot of the probabilistic findings, showing simulated estimates of cost difference against QALY difference between ERS and usual care, is provided in Figure 19. The scatterplot shows that all the simulations generated an improved effectiveness of ERS, but also a higher cost than usual care (i.e. all points were in the north-east quadrant of the cost-effectiveness plane). This reflects the relatively modest uncertainty around the cost of the intervention and assumptions about the distribution of uncertainty around the estimates of effect size.

FIGURE 19.

Cost-effectiveness plane showing the scatter plot of 10,000 Monte Carlo simulations for ERS compared with usual care.

The decision as to whether or not these findings can be considered cost-effective depends on the maximum amount decision-makers are willing to spend to obtain an additional unit of effectiveness (in this case, a QALY). This can be best presented in the form of a cost-effectiveness acceptability curve, as presented in Figure 20. At a threshold of £20,000 there is a 0.508 probability that ERS is cost-effective. This increases to 0.879 when a threshold of £30,000 is considered.

FIGURE 20.

Cost-effectiveness acceptability curve showing the probability of cost-effectiveness for ERS at varying levels of threshold.

Subgroup analysis of exercise referral schemes in individuals with pre-existing conditions

The remit of this HTA report was to examine the clinical effectiveness and cost-effectiveness of ERS in individuals with a pre-existing condition. The cost-effectiveness evidence reviewed in Chapter 4 captured relatively little existing evidence on such individuals. Rather, ERS was used to mitigate against unhealthy behaviours or risk factors for future conditions.

The aim of this section is to evaluate the cost-effectiveness of ERS in people with a diagnosed condition known to benefit from PA. We focused on the top three conditions (Table 49) that have been found to benefit most from increases in PA (BHFNC34); obesity, hypertension and depression (see Appendix 1, Figure 21, for full list).

Summary of the cost–utility analysis

Our analysis attempts to estimate the cost-effectiveness of ERS using a cost–utility analysis framework similar to that used in previous analyses (NICE 200676). Our base-case assumptions result in a favourable cost-effectiveness ratio of £20,876 per QALY gained from ERS compared with usual care. It should be acknowledged that our base-case estimate includes some optimistic assumptions with respect to cost and effectiveness. However, our deterministic and PSAs suggest that there is a low possibility of the ICER increasing above £30,000 when these assumptions are relaxed.

Analysis of ERS in groups of individuals with pre-existing conditions suggests that it may be more cost-effective in these groups, than in a sedentary population. ERS is frequently prescribed to individuals with risk factors for CVD. Our subgroup analysis includes populations with obesity and hypertension to reflect these individuals. In these groups, the cost-effectiveness of ERS falls to around £11,000 per QALY. In a population with depression, ERS cost-effectiveness is more favourable, generating an ICER of approximately £8000. Given the higher risk of developing the long-term illnesses considered in the model in these groups, it is not surprising that the subgroup analyses produce more favourable ICERs. This is an encouraging finding and suggests that it might be possible to target ERS to individuals with pre-existing conditions in which the pay-offs/impact may be higher. However, there remain some major uncertainties over whether or not the evidence used to populate the model, derived from the meta-analysis, is applicable to these groups. There may be good reason to believe that uptake, adherence and effectiveness might differ according to the characteristics of the recipients. Although we have attempted to adjust the model to take into account differences in the rate of long-term illnesses, no data were identified as part of the effectiveness review to allow for adjustment of the effect of ERS in different populations. There is a pressing need for better primary evidence to inform these uncertainties.

Although our cost-effectiveness estimates suggest that ERS is a cost-effective use of NHS resources, it should be noted that the individual-level lifetime QALY gains are relatively modest (< 0.01 in our base-case analysis). This estimate is predicated on the evidence of effectiveness derived from the meta-analysis presented earlier in this report. We believe that the meta-analysis has provided the most robust estimate to date of the effectiveness of ERS compared with usual care. However, it should be acknowledged that the cost-effectiveness analysis is attempting to capture lifetime benefits based on evidence of relatively modest effect sizes derived from short-term studies. Any such analysis inevitably involves some assumptions about the degree to which behaviour change is lasting and fails to consider other health behaviours that may impact on long-term outcomes. The result is that the cost-effectiveness analysis estimates that ERS has a modest lifetime cost and a marginal lifetime QALY gain. Even small changes in the source data used to populate the model, particularly evidence of effect size and cost, may lead to significant changes in the resulting ICER. This can best be illustrated through consideration of the net benefit calculation. If we value each QALY gained at £30,000 and accept that our analysis is generating a lifetime QALY gain of approximately 0.008 in most cases, then the value of the benefits generated in monetary terms is approximately £240, which exceeds the cost of the intervention. However, even a modest change in the lifetime QALY gain, to 0.07, would result in the costs exceeding the benefits, making the cost-effectiveness of ERS questionable.

Although sensitivity analysis has sought to address this point, it should be acknowledged that, in many cases, source data were derived from a single study (e.g. cost data from Isaacs et al. 61) and it was necessary to fit distributions to parameters to allow for PSA. Although every effort has been made to explore uncertainty, there is a possibility that the uncertainty around parameter estimates may be greater than predicted within our analysis, which would have a material impact on the ICER.

Although some caution should be taken in interpreting the findings, the authors would wish to emphasise that the estimates of cost-effectiveness generated are believed to be conservative. Our approach generates a partial analysis that considers only the impact of ERS on a number of morbidities known to be associated with PA. The impact on other morbidities was excluded owing to limitations in the available evidence. On this basis, our estimates of cost-effectiveness should be regarded as conservative, as we have made no attempt to quantify these benefits within our analysis.

Limitations of the analysis

The analysis had a number of limitations which should be acknowledged. First, we examine only the long-term impact of PA on selected morbidities. It was not possible to include other morbidities that may be affected by PA owing to uncertainty over the relationship between PA, incidence and quality-adjusted life expectancy. Nor does our model account for potential negative outcomes of PA, such as injuries. Although this may be an important determinant in taking up PA, particularly in the elderly, the evidence on injuries suggests that they are rare (Munro et al. 48), and they are not expected to significantly affect results when considered at a population level. Another set of limitations include assumptions relating to constant and independent risk of experiencing disease health states and age at onset of disease. These assumptions were derived from the NICE 2006 report76 and were meant to allow our analysis to be comparable with previous research. Although we recognised that these assumptions are limiting, their impact on the ICER, when investigated through sensitivity analysis, was considered minimal.

A number of other weaknesses in the model design were identified which were prioritised for further analysis. These include:

• the potential to capture the short-term improvements in QoL associated with PA (process benefits), which may be particularly important in certain groups, such as those who are prescribed PA for mental-health problems, such as depression

• the wide range of health benefits associated with increases in PA, including mental health, cancer and musculoskeletal conditions, which are currently excluded from the analysis.