Physical activity (PA) is associated with reduced risk for health conditions including coronary heart disease, cancers, diabetes, and stroke [1–3]. Despite interventions to increase PA in the English general population, the Health Survey for England (HSE) (2008) showed that only 39% of men and 29% of women in England are meeting the recommended level to be considered ‘physically active’ as defined by guidance from the Chief Medical Officer; only 6% of men and 4% of women, however, met the recommended level when PA was objectively measured [4].

Health Related Quality of Life (HRQoL), including the physical and mental well-being of an individual, is an important concept in health research and can help to inform decisions on prevention and treatment of ill health [5]. Estimates of the relationship between HRQoL and particular health states have served as inputs for economic evaluations intended to inform resource allocation decisions [6]. From a public health perspective a better understanding of how healthy lifestyles, such as uptake of PA, can influence HRQoL might help to inform policy intended to incentivise PA in the general population [5]. Evidence on the association between HRQoL and PA in the general population is limited as research, to date, has focused on specific interventions or populations with chronic conditions [7]. The dearth of evidence perhaps partly explains why economic evaluation of exercise interventions, to date, has rarely accounted for HRQoL gains directly attributable to increased exercise [8], so called process utility, and rather has focussed on the long-term effects of sustained PA on the incidence of chronic conditions.

A recent systematic review [7] showed that the few studies that have attempted to address this gap in knowledge, found a positive association between HRQoL and PA. The limited evidence base is, however, considered weak as the studies have methodological issues with respect to the measurement of PA [7, 9]. The studies mainly used subjective (self-reported) measures of PA without adequate validation. Some commentators argue that regardless of appropriate validity and reliability tests, the use of subjective measures of PA are subject to overestimation [10]. The few studies [11, 12] that used objective measures, however, relied on cardio-respiratory fitness, which unlike electronic devices (e.g. accelerometers) is not a direct measure of PA. As being a biological phenotype, cardio-respiratory fitness is influenced by both PA level and genetics, and the consequent interaction of both [7]. A notable exception is Hamer and Stamatakis [9] that used accelerometer data to explore the association between objectively assessed PA and subjective measures of well-being.

The present study is unique because it uses both objective and subjective measures of PA, and focuses on a broader measure of health status as it is the common endpoint in economic evaluation of public health interventions. The paper aims to explore the relationship between HRQoL and PA, and examines how this relationship differs across objective and subjective measures of PA, within the context of a large representative national survey from England. This study focuses on adults aged 40–60 years in order to: (a) relate the findings more directly to current evidence on exercise referral interventions in the UK as the literature centres on this age group; and (b) reflect the broader objectives of the programme of research exploring those interventions [8]. This research is part of a larger programme of research that explored the effectiveness and cost effectiveness of exercise referral schemes in the United Kingdom [8].

The results support the hypothesis that higher levels of PA are associated with better HRQoL. This relationship is consistent across different measures and types of PA. Participation in walking and sports and exercise are correlated with a modest effect on HRQoL. However, differences in the magnitude of HRQoL benefit associated with objective and subjective measures of PA are noticeable, with the former measure being associated with a relatively better HRQoL. An explanation for this could be that people tend to over-report their participation levels in PA [32], for example, 90 minutes of at least moderate intensive PA captured via a subjective measure might actually equate to a shorter duration or a less intensive period of exercise when measured objectively. The potential impact of this is that it yields a downward bias in econometric estimation with respect to subjective measures of PA, meaning that the link between HRQoL and PA is underestimated. This is further supported by descriptive statistics that indicate that people considered physically active via objective measure had slightly better mean HRQoL value (0.918) than individuals considered physically active via self-reports (0.916).

A key question is whether in order to achieve more accurate estimations, objective rather than subjective measures ought to be favoured. We are inclined towards a cautious approach, allowing for further corroboration of the findings because of a number of reasons. First, there was discrepancy in the reference periods of data collection for the two measures, subjective data was based on four weeks prior to survey date while objective data was based on a seven day period after the survey date. Second, the objective measure used herein, accelerometry, do not capture all types of PA with equivalent precision tending to underestimate activities like cycling and walking up stairs [4].

It is important that the limitations of the analysis are acknowledged. First, as a cross sectional study, the findings point to an association between HRQoL and PA and that no conclusions on causality can be drawn. Therefore, we cannot conclude whether participation in PA leads to higher HRQoL or improved HRQoL leads to a lifestyle (e.g. absence of limiting illnesses) more enabling for participation in PA. Nonetheless, the consistency of our findings across different types of PA strengthens the need for further research into the causal pathways between PA and HRQoL using longitudinal datasets. Second, the findings may not be generalisable to other age groups given the emphasis on adults aged 40–60 years. However, the findings are consistent with previous evidence on the relationship between PA and HRQoL [11, 12, 16]. Further confidence can be drawn from the findings because all regression models had good specification and fit.

The findings suggest that there is a relationship between PA and HRQoL. However, the nature of the study means that it is not possible to accurately dissect the nature of this relationship. It is important to consider the constituents of the HRQoL gains associated with PA captured in this study. Participation in PA generally tends to be associated with two main types of benefits i.e. physical benefits (reduced risk for ill-health conditions, improved fitness levels) and psychological benefits (e.g. mental simulation during participation, improved psychological health) [33, 34]. Given the cross sectional nature of our study and the fact that ill-health conditions were controlled for in the analysis, the benefits captured may more likely be psychological benefits. However, we are not able to determine the exact nature of these benefits. For example, the improvements in HRQoL associated with PA might be considered to be ‘process benefits’ that arise from engagement in PA. These might occur due to the process of participating (increased social interactions resulting from group participation or time spent outdoors), improved self-esteem (e.g. positive perceptions of competences and the physical-self) or biologic mechanisms (increased endorphin levels as a result of PA). Further research is needed to understand the relationship between PA and HRQoL and this may need to comprise both quantitative and qualitative methods.

Related to this, is the degree to which these benefits are lasting which also warrants further exploration as its crucial to how HRQoL gains are converted into outcomes (e.g. quality adjusted life years or ‘QALY’s’) in an economic evaluation. Sustaining PA levels is necessary if the risks of long-term conditions are to be modified. However, it remains unclear whether any process benefits associated with PA, as defined above, are sustained over time or whether these are associated with moving from a sedentary lifestyle to an active lifestyle, meaning individuals accrue a ‘one-off’ benefit to HRQoL. One might consider that HRQoL improvements arising from improved self-esteem are transitory whereas improvements resulting from biologic mechanisms might be sustained over time.

Notwithstanding these limitations, the findings here do indicate the potential for generating policy relevant information on promotion and evaluation of PA programmes. Programmes aimed at improving the uptake of PA might encourage participation rates by making people aware of the potential gains (which might occur sooner than later) associated with participation. In terms of economic evaluation, the findings here reinforce the need for investigation into the impact of the inclusion of such HRQoL gains. Previous analyses have suggested that including such benefits, alongside the longer-term benefits resulting from reduced incidence of long-term conditions, will lead to an improvement in the cost effectiveness of interventions designed to increase PA [35]. However, further research is required to understand the relationship between HRQoL improvements which might result directly from participation in PA as differentiated from improvements that result from sustained participation and reduced incidence of long-term conditions. Only by examining these relationships will we be in a position to accurately determine the cost effectiveness of interventions designed to promote PA.

Higher levels of PA are associated with better HRQoL. Using an objective measure of PA compared with subjective shows a relatively better HRQoL. Given the measurement errors associated with subjective measures, a key question is whether in order to achieve more accurate estimations of HRQoL, objective rather than subjective measures of PA ought to be favoured. We are inclined towards a cautious approach, allowing for further corroboration of the findings particularly because of discrepancy in the reference periods of data collection for the two PA measures.