Economic evaluation results

Parts of this chapter have been reproduced with permission from Willis et al.1 and Jago et al.2 These are Open Access articles distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: https://creativecommons.org/licenses/by/4.0/. The text below includes minor additions and formatting changes to the original text.

Introduction

This chapter provides an overview of the results relating to the economic analysis of the PLAN-A programme. We report the resource use and costs (both school and local authority) associated with setting up and delivering the PLAN-A intervention to Year 9 girls, as well as the outcomes of HRQoL (as measured using the EQ-5D-Y, KIDSCREEN-10 and mapped CHU-9D) and minutes of MVPA per weekday.

Intervention-costing results

Resources captured as part of the intervention cost include:

• Set-up costs for the PLAN-A programme. These comprised train-the-trainer costs and PLAN-A intervention consumables. In the main, set-up costs represent fixed costs and, therefore, the mean cost per school and pupil would decrease in proportion with the number of schools that receive peer supporter training.
• Programme delivery costs related to individual schools. These comprised staff costs (local authority and school staff) associated with the delivery of peer nomination and peer supporter training. Programme delivery costs (which, in the main, represent variable costs) were captured and presented at the level of the individual school. All 10 intervention schools contributed to the cost data collection.

Intervention set-up costs

Train-the-trainer costs

The train-the-trainers course was a 3-day event held in one location for all of the trainers. Two trainers (the lead trainer and one member of the research team) delivered the course to a group of six trainers, with a resultant event cost of £319.09 per school and £3.53 per pupil (where a pupil is defined as a girl on the register at the start of Year 9; Table 24).

Intervention consumables

Details on intervention consumables associated with the production of the PLAN-A resource kits were captured from invoices and entered on the PLAN-A trial budget database (Table 25). Intervention consumables (print and related resources) equate to £250.76 per school or £2.77 per pupil.

Intervention delivery costs

Local authority staff time

Staff time associated with the co-ordination and delivery of peer nomination and peer supporter training was valued. In the real world, it is assumed that these activities would be carried out by a local authority-employed delivery agent. In total, the local authority staff time (circa 58 hours, on average) and resource to deliver the intervention equated to £1651 per school and £18.26 per pupil (Table 26).

School staff time

Within schools, school staff resource varied by the grade of the staff member involved (i.e. leadership, PE or administration/support teacher). In addition, the PLAN-A school contact time varied. At the extremes, school 10 recorded very little (240 minutes) school staff contact with the intervention, whereas school 16 recorded 1275 minutes. Overall, school staff time equated to a mean of £595.88 per school and £6.59 per pupil (Table 27). Although PLAN-A was delivered in usual school hours and these costs do not represent additional budgetary costs for the school, it is still the case that the teacher’s input in other PE or school activities (on average, around 18 hours) will have been displaced by involvement in PLAN-A activities.

Total costs of intervention delivery

Table 28 presents the total cost of intervention delivery for each of the 10 intervention schools. The variable school staff costs (per hour) affected overall costs per school. There was variation in per-pupil costs between the schools, with costs ranging from £20.85 to £48.86. As the intervention involved high fixed costs, the cost per pupil was associated with the size of the school.

Table 29 shows how the mean intervention cost per school (see Table 28) is broken down. The bulk of the costs came from the peer supporter training days and the staff time used to support the intervention.

EQ-5D-Y VAS, KIDSCREEN-10 and CHU-9D scores

Overall, the percentage of girls on the register who completed the EQ-5D-Y VAS, KIDSCREEN-10 and CHU-9D instruments was 82%, 84% and 75%, respectively, at T0, and 74%, 75% and 74%, respectively, at T1. Of these, 72%, 75% and 73%, respectively, had no missing data at either T0 or T1.

Quality-of-life scores (measured using the EQ-5D-Y VAS, KIDSCREEN-10 raw scores and T-scores, and the mapped CHU-9D) deteriorated slightly from T0 to T1 in both the intervention and control arm (Table 30). Appendix 4 presents the mapped data for those with complete CHU-9D questionnaires by time point and trial arm.

Inspection of the profile responses of the EQ-5D-Y for each dimension of the instrument showed a similar pattern in both arms (Figures 4–8). A large percentage of girls (around 45%) at both time points and in both the intervention and control arms reported some or a lot of problems with pain or discomfort. The majority of girls (87–88%) in both groups reported that they were a bit or very worried, sad or unhappy; this proportion increased slightly at T1. It was unclear whether this was an age-related phenomenon or indicative of factors not measured in this trial, but consistency between arms suggests it was not related to the intervention. The intervention and control groups reported mean baseline CHU-9D index scores of 0.83; these scores declined by T1 (see Table 30).

FIGURE 4 Profile of the population: percentage reporting problems with mobility. (Figure not reproduced here; see the original article.)

FIGURE 5 Profile of the population: percentage reporting problems with looking after themselves. (Figure not reproduced here; see the original article.)

FIGURE 6 Profile of the population: percentage reporting problems with doing usual activities. (Figure not reproduced here; see the original article.)

FIGURE 7 Profile of the population: percentage reporting problems with pain or discomfort. (Figure not reproduced here; see the original article.)

FIGURE 8 Profile of the population: percentage reporting problems with feeling worried, sad or unhappy. (Figure not reproduced here; see the original article.)

Primary economic analysis

Cost per unit change in moderate to vigorous physical activity and cost per 5-minute increase in moderate to vigorous physical activity

The bootstrapped estimate of mean incremental costs was £30.50 and the mean incremental benefit was –3.54 minutes of MVPA. The mean estimate of the ICER, calculated as the ratio of the bootstrapped cost-effect replicates, was –£37.24, with the negative ICER, in this instance, indicating that the intervention is dominated (i.e. more costly and less beneficial than the control). The majority of the bootstrapped replications were in the north-west quadrant of the CEP (Figure 9), indicating that the intervention is probably more costly and less effective at improving weekday MVPA than the control. The scatterplot crosses the y-axis, indicating that there is, however, a high degree of uncertainty surrounding the effect on weekday MVPA. The CEAC (Figure 10) shows that the probability that the intervention was cost-effective is low over a range of willingness-to-pay thresholds for a unit change in MVPA. In secondary analysis, estimating the cost per 5-minute increase in weekday MVPA, the estimate of the ICER was –£186.22.

FIGURE 9 Mean cost and weekday MVPA and associated uncertainty (1000 bootstrap replications). (Figure not reproduced here; see the original article.)

FIGURE 10 The CEAC over a range of hypothetical willingness-to-pay thresholds for MVPA. (Figure not reproduced here; see the original article.)

Cost–consequences analysis

As has already been reported (see Table 17), pupils in the intervention and control arms achieved, on average, 45 and 49 minutes of MVPA per weekday, respectively. After adjustment, the difference between arms was estimated as –2.84 (95% CI –5.94 to 0.25; p = 0.071) minutes, a trend suggesting that those in the intervention arm performed fewer minutes of weekday MVPA than those in the control group, on average.

There was no evidence of a statistically significant difference in incremental effects between arms for (the mapped) CHU-9D or QALYs (Table 31). In unadjusted analyses, there was some evidence that the KIDSCREEN-10 and EQ-5D-Y VAS scores were better in the intervention group (see Table 31). However, differences were small and consistent with no effect of the intervention in adjusted analyses.

Cost per quality-adjusted life-year

In the exploratory cost–utility analysis using the mapped CHU-9D scores, the point estimate of the probabilistic ICER was £35,748 per QALY, which is above the £20,000-per-QALY threshold conventionally applied in the UK. The bootstrapped replications of the cost–effect pairs in Figure 11 show a similar pattern to the primary economic analysis; however, the replications are split more evenly between the north-west (i.e. more costly, less effective) and north-east (i.e. more costly, more effective) quadrants of the CEP. Figure 12 identifies that at conventional willingness-to-pay threshold values of £20,000 and £30,000 per QALY, the intervention has a 48% and 51% chance of being cost-effective, respectively. At lower thresholds, the intervention is much less likely to be cost-effective.

FIGURE 11 Mean cost and QALY and associated uncertainty (1000 bootstrap replications). (Figure not reproduced here; see the original article.)

FIGURE 12 The CEAC over a range of hypothetical willingness-to-pay thresholds. (Figure not reproduced here; see the original article.)

Consideration of long-term consequences

Our published health economics analysis plan96 considered long-term consequences on the basis of the intervention meeting two conditions: (1) evidence of promise and (2) availability of a model linking changes in adolescent physical activity to long-term health outcomes. Evidence of promise required that the point estimate of MVPA was positive and the 95% CI included the possibility of a meaningful (5 minutes of weekday MVPA) positive intervention effect. As this criterion was not met and there was no available model linking changes in adolescent physical activity and health outcomes, an extrapolation model is not presented.