Summary of exercise referral scheme effectiveness

Seven randomised controlled trials (RCTs; UK, n = 5; non-UK, n = 2) met the inclusion criteria, recruiting a total of 3030 participants (1391 randomised to ERS). Five studies compared ERS with usual care, two studies compared ERS with an alternative PA intervention (walking or motivational counselling programme) and one study compared ERS with ERS plus a self-determination theory (SDT) intervention. Studies were judged to have a moderate-to-low risk of bias. The most consistently reported outcome was self-reported PA. In an intention-to-treat analysis, compared with usual care, there was weak evidence of an increase in the number of ERS participants who achieved 90–150 minutes of at least moderate-intensity PA per week at 6–12 months’ follow-up [pooled relative risk (RR) 1.11, 95% confidence interval 0.99 to 1.25]. There was no consistent evidence to support a difference between ERS and usual care in the duration of moderate/vigorous-intensity and total PA, physical fitness, blood pressure, serum lipids, glycaemic control, obesity indices (body weight, body mass index and per cent fat), respiratory function, psychological well-being (perception of self-worth, symptoms of depression or anxiety) or HRQoL. There were no differences in PA or other outcomes in ERS versus alternative PA interventions or versus ERS plus a self-determination intervention. None of the included trials separately reported outcomes in individuals with medical diagnoses.

Summary of predictors of exercise referral scheme uptake and adherence

Fourteen observational studies and five RCTs provided a numerical assessment of ERS uptake and adherence (UK, n = 16; non-UK, n = 3). There was considerable evidence of variation in levels of both ERS uptake (35–100% of people attending the first ERS induction visit) and adherence to ERS (12–82% of people taking up ERS completing the programme). ERS uptake levels were generally higher in RCTs (79%) than in observational studies (62%), with no clear difference in adherence between different study designs (37% vs 48%). Women and older people appeared to be more likely to take up ERS. Furthermore, while older people were also more likely to adhere, women were less likely to adhere than men. There was little evidence to be able to judge the influence of participant psychosocial or programme-level factors on ERS uptake or adherence. The majority of the 10 included qualitative studies highlighted participants’ perception of a range of short-term physical and psychosocial benefits associated with ERS.

Summary of exercise referral scheme cost-effectiveness

Four previous economic evaluations (UK, n = 3; non-UK, n = 1) assessing the cost-effectiveness of ERS were identified – three trial-based economic evaluations and one model-based analysis. Broadly, the evidence base suggested that ERS was a cost-effective intervention in sedentary populations without a medical diagnosis.

Indicative incremental cost per QALY estimates for ERS for various scenarios were based on de novo model-based economic evaluation. Compared with usual care, the mean incremental cost for ERS was £169 and the mean incremental QALY was 0.008, with the base-case incremental cost-effectiveness ratio (ICER) for ERS at £20,876 per QALY in sedentary individuals without a diagnosed medical condition and £14,618 per QALY in sedentary obese individuals, £12,834 per QALY in sedentary hypertensive patients, and £8414 per QALY for sedentary individuals with depression; however, findings report small incremental costs and QALYs, and ICERs were therefore highly sensitive to plausible variations in the RR for change in PA and cost of ERS.

Strengths, limitations, uncertainties of the analysis

Our electronic database searches were restricted to controlled trials, to examine the highest level of evidence for effectiveness, with ERS studies carefully selected on the basis that there was clear evidence of referral by a primary-care health professional to a third-party exercise provider. We extended the scope of this report to undertake a review of quantitative and qualitative literature so as to better understand the potential predictors of ERS uptake and adherence. However, we did not incorporate formal methods of qualitative synthesis such as meta-ethnography. A particular strength of our cost-effectiveness analysis was the further development of the economic model originally used in the NICE evaluation of primary care-based exercise interventions. These further developments included the incorporation of epidemiological data linking PA and the future risk of clinical outcomes in specific diagnoses groups (i.e. obesity, hypertension and depression). For the purposes of generating a cost per QALY for people with a specific medical diagnosis, we assumed that the same benefit in terms of PA gains in those populations as sedentary ‘at-risk’ individuals.

Because of limitations and gaps in the evidence base there remain several key uncertainties regarding the effectiveness of ERS. These include (1) the certainty in the improvement in short-term PA seen in sedentary individuals without a medical diagnosis; (2) the impact of ERS in people with a medical diagnoses; (3) whether or not ERS consistently affect clinical outcomes such as blood pressure and serum lipids; and (4) whether or not the potential small gains in short-term self-reported PA with ERS are maintained over the longer term. The cost-effectiveness for ERS is uncertain because of the limitations and gaps in the clinical effectiveness evidence base. Sensitivity analyses show that the cost per QALY associated with ERS can change markedly, with plausible changes in model effectiveness and cost inputs, which means that robust evidence on whether or not ERS are likely to be cost-effective cannot currently be provided.

Implications for service provision

In 2006, NICE commented that there was insufficient evidence for ERS and recommended that the NHS should only make ERS available as part of a controlled trial. Although we have identified four additional trials since the NICE review, there remains very limited support for the potential role of ERS for impacting on PA and, consequently, public health. Arguably, such an uncertain impact provides a case for the disinvestment in ERS. However, little evidence was found of how the ERS intervention sought to develop a sustainable active lifestyle in participants, as recommended in the NHS National Quality Assurance Framework. Although ERS programmes in our review aimed to increase medium- to long-term PA, they were typically based on only a 10- to 12-week leisure centre-based period intervention. With the exception of one trial (Jolly K, Duda JL, Daley A, Ntoumanis N, Eves F, Rouse P, et al. An evaluation of the Birmingham exercise on prescription service: standard provision and a self-determination focused arm. Final Report; 2009), there was minimal reference to health behaviour change techniques and theories that typically underpin interventions to promote an increase in daily PA.

Research priorities

• Randomised controlled trials assessing the clinical effectiveness and cost-effectiveness of ERS in disease groups that might benefit from PA. In addition, RCTs should seek to incorporate hard to reach populations (e.g. ethnic minorities) that are traditionally not represented in trials.
• Such RCTs should be better reported, include long-term data on the clinical effectiveness of ERS and the sustainability of PA change, incorporate objective measures of PA (e.g. accelerometers) and health outcomes (e.g. blood pressure, serum lipids) and incorporate parallel-process evaluations to better understand the mediators and barriers to behaviour change.
• Exercise referral scheme programmes vary in their procedures and this may impact on uptake and adherence. Future trials should, therefore, be designed to better understand the contribution of different programme components (e.g. level of staff training) to the clinical effectiveness and cost-effectiveness of ERS.
• Head-to-head RCTs comparing the clinical effectiveness and cost-effectiveness of different models of primary-care interventions aimed at promoting PA.
• Further quantitative and qualitative studies are needed to determine the moderators of uptake and adherence to ERS.
• Theory-driven interventions should be developed to complement ERS to foster long-term change in PA, and evaluated to enhance our understanding of mediators and processes of behaviour change (e.g. SDT, motivational interviewing).
• The development of improved approaches to modelling the cost-effectiveness of ERS, capturing the potential impact on a wide range of health outcomes.