Health economics evaluation

Background

In this project, aflibercept was evaluated as an alternative treatment to PRP in PDR without macular oedema. The clinical efficacy results showed that aflibercept is superior to PRP in terms of visual acuity change at 52 weeks. In this chapter, we report the economic evaluation of aflibercept as a treatment option for PDR.

Objective

Drawing on the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) checklist,68 we report cost-effectiveness of intravitreal aflibercept compared with PRP.

Methods

Economic evaluation

From a public sector multiagency perspective that covers health and social care services, follow-up at 52 weeks was based on data collected from the CLARITY study, a Phase IIb, single-blind, non-inferiority trial on adults (aged ≥ 18 years) with type 1 or 2 diabetes mellitus and previously untreated or post-laser treated active PDR. Study participants were recruited from 22 UK ophthalmic centres.

Cost of intervention (aflibercept) and comparator (panretinal photocoagulation)

We included both the purchase price of aflibercept and the injection procedure cost. PRP was costed using the published national mean unit cost.46 Aflibercept injection treatment in the intervention arm and PRP in the comparator arm were costed using published national unit costs from the Department of Health46 and BNF47 at 2016 price year.

Costing service use

We explored patterns of hospital- and community-based service use over the 52-week follow-up study period, using a CSRI.69 National unit costs from published sources at 2016 price year were used to calculate health and social care service use costs.46,48,70,71 We bootstrapped differences in cost to produce a 95% CI around these differences.48

Measurement of outcomes

In our primary cost-effectiveness analysis, we used BCVA as the measure of effectiveness. Change in BCVA score from baseline was calculated for intervention and comparator groups and the difference in change in BCVA score between groups was assessed.

For HRQoL, we collected EQ-5D-3L data at baseline and at 52 weeks post baseline. The EQ-5D-3L index scores were calculated at baseline and 52 weeks for both the intervention and comparator groups. Potential QALY gains were then calculated using the standard ‘area under the curve’ (AUC) method.72,73 QALY is a generic preference-based measure weighing the quantity of life (i.e. survival or the number of additional life-years) by the quality of life of participants, experienced at baseline and 52-week follow-up.

From a methodological perspective, we compared the performance of the EQ-5D-3L (a generic, preference-based HRQoL measure) with the vision-specific (non-preference-based) HRQoL measures (NEI-VFQ-25 and RetDQoL). We included the ICECAP-A measure as an alternative to EQ-5D-3L, focusing on capability to see how this measure compared in terms of sensitivity in this patient group.74 We did not collect information on activities of daily living (ADL), which would have told us something about how treatment of this condition with aflibercept versus PRP affects home and work life.

Cost-effectiveness analysis

From a public sector multiagency perspective, which covers health and social care services, we assessed the cost-effectiveness of intravitreal aflibercept compared with conventional PRP. The incremental costs and effects of alternative arms were compared in the primary cost-effectiveness analysis with effectiveness measured in terms of BCVA. Published sources of national unit costs at 2016 price year were used to calculate the total cost of health and social care service use and ophthalmic-related drugs use over 52 weeks by patients in each arm of the study. We employed ‘bootstrapping’ to overcome the skewed data and produced CEACs to quantify uncertainty. For the cost-effectiveness analysis and cost–utility analysis, difference in cost and effects (change in BCVA and QALY) between arms of the study were adjusted for differences in baseline cost and effects.75 We undertook this adjustment by applying linear regression in IBM SPSS Statistics 22 (IBM Corporation, Armonk, NY, USA) with baseline cost and trial arm as the only covariates for cost adjustment, baseline BCVA and trial arm as the only covariates for change in BCVA adjustment, and baseline EQ-5D-3L index score and trial arm as the only covariates for QALY adjustment.76

Secondary exploratory cost–utility analysis

We undertook a secondary exploratory cost–utility analysis, that is, a cost per QALY analysis. Evidence is mixed but points to the EQ-5D-3L not being sufficiently sensitive to be useful in studies of visual impairment.44,45

Sensitivity analysis

We used sensitivity analysis to explore whether or not the estimated effect (change in BCVA score) and cost of aflibercept injection treatment relative to PRP laser treatment were sensitive to the key variable factor of our analysis – the pricing of aflibercept. This is most important variable factor because the NHS PAS can make the drug available at a range of prices.

Subgroup analysis

We undertook a subgroup analysis to explore the potential effect of heterogeneity defined in terms of DMO. The subgroup analysis was performed by the presence of DMO at baseline (no DMO in both eyes vs. DMO presence in at least one eye). The incremental costs and effects of alternative arms were reported within each of the categories of the subgroup variable, and the ICER (i.e. cost per change in BCVA) was calculated for both of these categories of the subgroup variable.

Results

Economic sample

Economic evaluation was undertaken on 202 participants (101 per arm) with complete cost and outcome data. This represents 96.7% of the clinical sample included in primary outcome ITT analysis. The economic sample was made up of male (n = 134) and female (n = 68) participants with an overall mean age of 51.3 years (SD 13.4 years). There was no statistically significant difference in the EQ-5D-3L index score at baseline (p > 0.05) or ICECAP-A score at baseline (p > 0.05).

Cost of intervention (aflibercept) and comparator (panretinal photocoagulation)

The list price for aflibercept was £81645 with an administration cost of £182 per injection procedure (Department of Health and Social Care, 2016). PRP laser treatment costs £131 per procedure.44 All participants in both groups received FFA, which costs £117 per angiogram.45 A full list of hospital- and community-based and ophthalmic related medication unit costs and their sources are shown in Appendix 3. All unit costs were at 2016 price year.

Frequency and cost of service use over 52 weeks

Table 23 shows the frequency of contacts with primary and secondary care health services and other services used by 202 participants in the intervention and comparator arms of the CLARITY trial over the 52-week study period.

Calculation of total mean service use costs over 52 weeks

Table 24 shows the mean cost of health and social care service use and ophthalmic related drug use over 52 weeks of study follow-up. The price year of all costs was 2016. No discount rate was applied as the study ran for 12 months.

Total mean costs of service use included primary care consultations, secondary care consultations and other community-based services, for example social work. Table 24 shows that the mean adjusted total cost (including ophthalmic-related drug cost) per participant was £7775.89 (SD £985.94) for the aflibercept group and £2301.31 (SD £948.65) for the PRP group. The difference in mean total cost between the two arms was £5474.57 (bootstrapped 95% CI £5210.79 to £5749.82). Primary care and other community-based costs accounted for a very small proportion of total mean costs (2.7% in the intervention group and 7.7% in the comparator group). Hospital-based costs, specifically the cost of aflibercept or PRP, accounted for the majority of total service use costs.

Measurement of outcome

Table 25 shows mean and SD scores for BCVA (our primary measure of effectiveness) in the intervention and comparator groups over 52 weeks. The far-right column shows the difference in mean adjusted change scores between arms with the bootstrapped 95% CI (5000 replications). We observed an adjusted 3.93-point difference on the BCVA (bootstrapped 95% CI 3.84 points to 4.02 points) between study arms in our economic sample. This represents a statistically significant difference between groups and is exactly the same as the difference observed in the ITT clinical sample analysis.

Table 26 shows the mean and SDs for EQ-5D-3L index scores and QALYs over 52 weeks in the intervention and comparator arms. At baseline, there was no statistically significant difference in mean EQ-5D-3L index scores between intervention and comparator arms (p = 0.774 > 0.05). Compared with UK national population norms, our EQ-5D-3L index scores at baseline were just a little lower than the population mean of 0.85 with similar SD.77 The far-right column shows a difference in mean incremental adjusted QALY of –0.022 between groups (using AUC methods) with the bootstrapped 95% CI (–0.080 to 0.034) and this difference is not statistically significant. We undertook a secondary exploratory cost–utility analysis, that is, a cost-per-QALY analysis. The results are shown in Table 26. The results speak for themselves; a mean adjusted cost difference of £5475 is divided by an extremely small and not statistically significant mean adjusted QALY difference of –0.022. This yields a cost per QALY of –£252,827, in which we do not have much confidence. Given that a positive significant difference was observed in the BCVA for the intervention group, we interpreted this as the EQ-5D-3L not being sufficiently sensitive in this context.

Table 27 shows the mean and SDs for ICECAP-A capability index scores and change in mean ICECAP-A index score over 52 weeks in the intervention and comparator arms. We observed no statistically significant change from baseline scores between groups using the ICECAP-A measure of capability as an alternative to QALYs as shown in the far-right column of Table 27.

Vision specific quality-of-life measures

As shown in the far-right columns of Tables 28 and 29, we observed no statistically significant difference in mean scores for NEI-VFQ-25 and RetDQoL between arms over 52 weeks. Table 28 shows the mean NEI-VFQ-25 composite scores, change in mean NEI-VFQ-25 composite score between study time points and difference in mean change scores between groups over 52 weeks. Table 29 shows mean RetDQoL scores for overview items 1 and 2 and AWI score, change in mean score between study time points and difference in mean change scores between groups over 52 weeks.

Cost-effectiveness analysis

In our primary cost-effectiveness analysis using BCVA as our measure of outcome, we divided the difference in mean adjusted total cost between arms by the difference in mean adjusted change in BCVA between arms to gain an ICER of £1392.99 for aflibercept as compared with PRP laser treatment. Figure 11 is a cost-effectiveness plane, a scatterplot of the joint distribution of incremental cost and effect (BCVA) between the two arms. Figure 11 shows the 5000 bootstrapped replications used to generate a 95% CI around our mean point estimate ICER. All bootstrapped points appear in the north-east quadrant (more effective but more costly) of our cost-effectiveness plane. Accompanying this plane is a CEAC (Figure 12) that shows that at the threshold of £1400, our hypothetical societal willingness to pay, there is a probability of 56.60% of aflibercept being cost-effective at its list price of £816.

FIGURE 11 Cost-effectiveness plane: using BCVA as the measure of effectiveness. (Figure not reproduced here; see the original article.)

FIGURE 12 Cost-effectiveness acceptability curve. (Figure not reproduced here; see the original article.)

Sensitivity analysis

The most important variable factor is the pricing of aflibercept. The NHS PAS could make the drug available at a range of prices to the NHS. In Table 30 we show how the ICER and probability of cost-effectiveness vary as a discount is applied to the list price of aflibercept. We have used a hypothetical societal willingness-to-pay threshold of £1400 to mimic the NICE threshold used in cost-per-QALY calculations. If society is willing to pay £1400 for an additional 1-point improvement in BCVA, then aflibercept has a 56.60% probability of being cost-effective at the list price of £816. From 20% through to 100% PAS, results showed 100% probability of aflibercept being cost-effective at the hypothetical societal willingness-to-pay threshold of £1400.

Subgroup analysis

We performed a subgroup analysis by the presence of DMO at baseline [no DMO in both eyes at baseline, n = 126, vs. DMO in at least one eye (in this study, this applies to fellow eye only as DMO was an exclusion in the study eye) at baseline, n = 76]. Table 31 shows that the incremental costs and effects of alternative arms were £5552.25 (bootstrapped 95% CI £5217.24 to £5873.87) and 3.98 (bootstrapped 95% CI 3.87 to 4.09) for the ‘no DMO in both eyes at baseline’ group, and £5351.48 (bootstrapped 95% CI £4894.25 to £5805.31) and 3.86 (bootstrapped 95% CI 3.70 to 4.02) for the ‘DMO in at least one eye at baseline’ group. This yields an ICER of £1394.27 per change in BCVA score (bootstrapped 95% CI £1305.87 to £1484.50) and £1387.45 per change in BCVA score (bootstrapped 95% CI £1395.86 to £1441.65) for the ‘no DMO in both eyes at baseline’ group and the ‘DMO in at least one eye at baseline’ group, respectively. The subgroup analysis showed there to be no meaningful difference in the cost per change in BCVA when subgroup analysis of this kind was undertaken.

Table 31 summarises our health economics findings. All analyses used bootstrapping with 5000 replications. In the far-right column of Table 31, we show how the probability that aflibercept is cost-effective increases when society is willing to pay more to achieve a 1-point improvement in BCVA.