Active children through incentive vouchers – evaluation (ACTIVE): a mixed-method feasibility study

Background Adolescents face many barriers to physical activity, demonstrated by the decline in physical activity levels in teenage populations. This study aimed to assess the feasibility of overcoming such barriers via the implementation of an activity-promoting voucher scheme to teenagers in deprived areas. Methods All Year 9 pupils (n = 115; 13.3 ± 0.48 years; 51 % boys) from one secondary school in Wales (UK) participated. Participants received £25 of activity vouchers every month for six months for physical activity or sporting equipment. Focus groups (n = 7), with 43 pupils, and qualitative interviews with teachers (n = 2) were conducted to assess feasibility, in addition to a process evaluation utilising the RE-AIM framework. Quantitative outcomes at baseline, five months (during intervention) and twelve months (follow-up) included: physical activity (accelerometer), aerobic fitness (12 min Cooper run) and self-reported activity (PAQ-A). Motivation to exercise (BREQ-2) was measured three months post-baseline and at follow-up. Results Qualitative findings showed that vouchers encouraged friends to socialise through activity, provided opportunities to access local activities that pupils normally could not afford, and engaged both those interested and disinterested in physical education. Improvements in weekend moderate-to-vigorous physical activity and reductions in sedentary behaviour were observed in both sexes. Boys’ fitness significantly improved during the voucher scheme. ‘Non-active’ pupils (those not meeting recommended guidelines of 60 mins∙day−1) and those with higher motivation to exercise had higher voucher use. Conclusions Adolescents, teachers and activity providers supported the voucher scheme and felt the vouchers enabled deprived adolescents to access more physical activity opportunities. Voucher usage was associated with improved attitudes to physical activity, increased socialisation with friends and improved fitness and physical activity; presenting interesting avenues for further exploration in a larger intervention trial. Electronic supplementary material The online version of this article (doi:10.1186/s12889-016-3381-6) contains supplementary material, which is available to authorized users.

Baseline PA (objectively-measured and self-reported) and aerobic fitness measures were obtained one month prior to the intervention, and were repeated during the fifth month of the intervention period and again after six months of follow-up post intervention. Motivation to exercise was measured mid-intervention and again at follow-up ( Figure 1).

Data Analysis
Accelerometry data were downloaded and converted to 60-second epoch .csv files using GENEActiv PC software version 2.1. These 60-second epoch data files were entered into an open source Excel macro (version2; Activinsights Ltd.), designed and validated for 60-second epochs, in order to classify non-wear and sleep time, as described by Metcalf and colleagues [9]. KineSoft software (version 3.3.75; KineSoft, Loughborough, UK) was then used to produce PA intensities (min.day -1 ) for each participant-day following procedures similar to those described elsewhere [10]. The widely used wear time criteria of 600 minutes on any three days was applied [11]. In addition, validated cut-points were used to classify sedentary behaviour, light physical activity (LPA) and moderate-to-vigorous physical activity (MVPA) [1].
Paired t-tests were conducted to investigate the differences from baseline to postintervention and follow-up for PA and fitness and similarly for motivation to exercise at two measurement time points: mid-intervention and six-month follow-up. Paired ttests were used instead of repeated measures ANOVA due to unequal numbers of observations between time-points, selection bias and loss of power if observations are required at all three time-points. Additionally, the assumption that compound variance would not differ in this instance could not be guaranteed. Preliminary analyses to ensure normal distribution of data were completed prior to all further analyses. Descriptive statistics (frequencies/percentages) were used to describe levels of engagement with the scheme and to gain insight into voucher usage. STATA V.12.1 (STATA, Texas, USA) was used for all statistical analyses and statistical significance was set at p<0.05 throughout.

Physical Activity and Fitness
Objectively assessed PA showed a marginal increase in MVPA during the voucher scheme (Table 4). When stratified by day of the week, weekend MVPA showed a significant increase during intervention in both sexes (Table 5). Similarly, LPA increased marginally over the six month period (Table 4), though this was only shown to be significant during weekdays. These increases were not maintained when vouchers were removed. A significant decrease in sedentary time was reported overall during the intervention, though this was not sustained twelve months post-baseline.
The intervention was associated with a significant improvement in fitness at postintervention (Table 4). Stratification by sex showed boys' fitness improved significantly, whereas girls' improved only marginally. These improvements were not sustained for boys at the twelve month follow-up and returned to baseline fitness levels after the vouchers were stopped. However, girls' fitness remained comparable to that found directly after the intervention.
The fitter, more active participants attended the baseline tests (n=87) and a greater proportion of those less fit attended at the end of the intervention (n=95). Fitness at baseline and post-intervention in the unfit pupils (defined as <1600m (girls)

Factors affecting voucher usage
The amount of time spent in sedentary pursuits at baseline (classified as sedentary minutes per day>median (667.4 min.day -1 )) had no effect on voucher usage. However, being classified as 'non-active' at baseline (MVPA<60min.day -1 ) was significantly associated with higher voucher use than 'active' pupils (21 ± 9.6 & 12 ± 10.8 vouchers, respectively, p=0.007). Pupils who were of average fitness or above at baseline (CRT score >1600m (girls) and CRT score >2200m (boys) [12], n=61) showed no difference in voucher usage compared to those below average fitness (n=54), (13 ±10.7 & 13 ±11.5 vouchers used, respectively, p=0.9). No significant differences with sex or free school meal eligibility were observed.
Voucher usage was also significantly associated with motivation scores, suggesting that those who were more motivated used more vouchers (regression coefficient 0.71 (p=0.001) using motivation as a continuous variable; most motivated 50% used 15 vouchers on average; less motivated 50% used 10 vouchers).