M.3.1. Base case
This analysis found that, given a NICE willingness-to-pay threshold of £20,000 per QALY gained, the most cost-effective strategy is likely to be one of starting with twice daily potent corticosteroid and moving to concurrent potent corticosteroid and vitamin D analogue and then twice daily coal tar. This strategy was also the least costly strategy among the 118 modelled. Base case results for non-dominated and non-extendedly dominated strategies are presented in .
Incremental analysis of base case results – psoriasis of trunk and limbs.
Results showed that starting with concurrent potent corticosteroid and vitamin D analogue and switching to twice daily potent corticosteroid and then twice daily coal tar is £9 more costly over 1 year and only produces 0.00041 more QALYs than the least costly strategy mentioned above. This gives it an incremental cost-effectiveness ratio (ICER) of £22,658 which is just above the NICE £20,000 per QALY threshold.
The most effective strategy (once daily TCF then twice daily potent corticosteroid then twice daily coal tar) costs an additional £192 per year compared to the next most costly non-dominated strategy (concurrent steroid and vitamin D then twice daily potent steroid then twice daily coal tar), yet produces just 0.00107 additional QALYs for an ICER of over £179,000. Based on the results of this model, it appears that starting with once daily TCF, although most effective, is very unlikely to be cost-effective.
Mean costs and QALYs and their respective 95% confidence intervals for all strategies, ranked in order of mean net benefits at a £20,000 per QALY threshold, are presented in . These show that a strategy of using vehicle or emollient with no active agent only was the most costly and least effective, largely driven by the cost of referrals and specialist management for non-responders. Strategies that included once or twice daily vitamin D were not cost-effective regardless of where they were included in the sequence. This is largely due to their relatively low rank in terms of effectiveness and their relatively high acquisition cost. Strategies that included dithranol were also all dominated, that is more costly and less effective than alternatives. Finally, strategies in which patients were referred after non-response to only 2 topicals were all dominated, thus not cost effective.
Mean total costs and QALYs for all modelled comparators.
A breakdown of total costs by type of resource use (i.e. topicals, GP visits, outpatient consultations, phototherapy) is presented for all modelled strategies in . Note that these costs were produced by a deterministic run of the model and therefore may not match exactly the total costs presented from the probabilistic analysis in ; however, they are very similar. Disaggregation of costs allows one to observe what part of a given strategy is driving the majority of total cost. Strategies that are less effective tend to have higher downstream costs driven by visits to the GP and referrals for specialist review and/or phototherapy. Strategies that are very effective are likely to have lower downstream costs, but potentially higher drug costs. Based on this disaggregation, it becomes clear that strategies with TCF product or vitamin D analogue have relatively high topical costs, some of which are offset by reduced downstream costs in terms of consultations with specialists and courses of phototherapy. Strategies with potent corticosteroids offered alone or in combination with vitamin D analogue (concurrent therapy) show similar downstream costs as strategies involving TCF product, but because their acquisition cost is dramatically lower, the overall total cost is much lower.
Disaggregated total costs by items of resource use.
The probabilistic analysis indicates that there is a great deal of uncertainty as to which sequence is optimal (i.e. most cost effective). There appears to be very little difference between initial potent corticosteroid followed by concurrent corticosteroid and vitamin D and vice versa, with the difference in their net monetary benefits (NMB) being only £1 (£16,748 and £16,747 respectively) and both having an equal probability of being optimal at a £20,000 willingness to pay threshold. Generally, it looks as though a strategy of starting with either potent corticosteroids or concurrent treatment with potent corticosteroid and vitamin D analogue is most likely to be cost-effective, whereas starting with once daily TCF product is very unlikely to be cost-effective.
M.3.2. Sensitivity analyses
A series of sensitivity analyses suggested that the conclusions from the base case are somewhat sensitive to changes in some parameters and/or assumptions.
M.3.2.1. Treatment effects
The network meta-analysis of topical therapies was performed for two response outcomes: investigator assessed global improvement (IAGI) and patient assessed global improvement (PAGI). The economic evaluation used the investigator assessed outcome in the base case, largely because there was more data from the randomised evidence reported for this outcome. In a sensitivity analysis, treatment effects from the network meta-analysis of patient reported outcome was used. Results of this sensitivity analysis are presented in .
Incremental analysis of sensitivity analysis using patient-reported outcome (PAGI).
Results of the analysis using patient reported outcomes indicates that starting treatment with once daily potent corticosteroids, moving on the concurrent treatment if that fails and then trying twice daily vitamin D analogue is likely to be both the least costly and most cost-effective strategy given a threshold of £20,000 per QALY gained. Initial treatment with concurrent potent corticosteroid and vitamin D analogue appears less cost-effective using patient reported outcomes than physician reported outcomes, unlikely to be cost-effective at thresholds less than £100,000. Once daily TCF product, first or second line in a sequence, still looks to generate additional benefits (QALYs), but at additional costs unlikely to be considered good value for NHS resource (ICERs upwards of £115,000 per QALY gained).
The base case network meta-analysis of physician/investigator assessed response used in the base case cost-effectiveness analysis included all RCTs that met the inclusion criteria for the clinical review of direct evidence. The review of direct evidence was quite focused and as such did not include evidence for every possible pair wise comparison. In a sensitivity analysis of the network meta-analysis and thus the cost-effectiveness analysis, additional studies were included. For details on the particulars of these sensitivity analyses and what effect they had on the estimated treatment effects, see Appendix K.
When treatment effects were based on all relevant RCT data, the results of the base case changed only slightly. Twice daily potent corticosteroid followed by concurrent steroid and vitamin D analogue is still likely to be optimal for first and second line treatments. However, instead of twice daily coal representing the optimal third line topical, twice daily vitamin D analogue looks to be most cost-effective. This sensitivity analysis calls into question whether vitamin D or coal tar represents the better third line treatment option.
M.3.2.2. Variation in early versus late response
The base case assumed that patients would trial a given topical for up to 8 weeks and that some proportion of patients would be expected to respond by 4 weeks and discontinue treatment at that time. The remainder would carry on to 8 weeks, at which time non-responders would move on to the next topical in a sequence. The data defining the breakdown of early (at 4 weeks) vs late (at 8 weeks) responders was limited to two studies41,73 and GDG opinion and was thus very uncertain. Deterministic sensitivity analyses were performed around these parameters to observe the impact on the results.
First, an analysis was performed in which no one was expected to respond and discontinue treatment at 4 weeks (i.e. all responders require 8 weeks treatment). Compared to the results of the base case when all comparators are included, the rank order of strategies in terms of mean net benefits changed very little. The ICERs for strategies on the cost-effectiveness frontier (see ) increased relative to the base case, thus becoming less likely to be considered cost-effective.
Second, an analysis was performed in which all responders were assumed to respond by 4 weeks, with no one requiring an additional 4 weeks of treatment. The ICER for all strategies on the cost-effectiveness plane (see ) decreased relative to the base case, and now starting with concurrent therapy and moving to twice daily potent corticosteroids looks to be cost-effective at a £20,000 threshold compared to potent corticosteroids and then concurrent therapy. Initial treatment with once daily TCF product is still unlikely to be cost-effective, with an ICER of more than £140,000.
Finally, an analysis was performed in which a 4-week stopping rule was applied. In this scenario, responders were limited to those that have responded by week 4 (see ), and all other patients are assumed to move on to the next topical in the sequence (i.e. no one continues to 8 weeks of treatment with the same topical). Relative to the base case, the total costs for all strategies more than doubled as more patients were classified as non-responders and moved down the care pathway reaching referral to secondary care. Starting with concurrent therapy and then moving to twice daily potent corticosteroids was now the least costly strategy and most likely to be cost-effective. The ICER for once daily TCF product instead of concurrent therapy in this sequence decreased substantially relative to the base case (£174,000 to £94,000) but is still unlikely to be considered cost-effective at the NICE threshold.
M.3.2.3. Reduced adherence
There was some concern that issues of treatment adherence were inadequately captured in the model. The estimates of effect used in the base case were derived from randomised controlled trials which may represent the best case scenario for topical therapies. The GDG wished to explore how reduced adherence to twice daily treatments would affect the conclusions of the base case. In this scenario, 60% of patients being treated with twice daily topical were assumed to adhere to twice daily treatment whilst the remaining 40% of patients were assumed to apply the topical only once daily74. For concurrent therapy, the 40% were assumed to adhere to once daily potent corticosteroid treatment only. Efficacy of the twice daily treatments would thus be reduced compared to the base case estimates. To be conservative, no reductions in cost were assumed despite the fact that less topical would be used.
With adherence reduced, there is no change substantive change to the results of the base case. Total costs across all strategies increase slightly (average of £27 more) and benefits decreased very slightly (average of 0.0007 fewer QALYs), but the conclusions from the base case remain unchanged. The most cost-effective strategy, given a £20,000 per additional QALY threshold is still twice daily potent corticosteroid followed by concurrent therapy and then twice daily coal tar. To put concurrent therapy before twice daily potent corticosteroids has an ICER of £36,000 (up from £23,000 in base case) and to replace concurrent therapy with once daily TCF before steroids has an ICER of £76,609 (down from £174,545 in the base case).
M.3.2.4. Utility values
In the base case, the mean utility gain associated with achieving some level of improvement, but not clearance or near clearance was assumed to be 0.05. This value was based on a downward adjustment of a value used in a recent cost-utility analysis included in the health economic review. Bottomley and colleagues62 modelled a utility gain of 0.07 for non-responders compared to baseline. To see what effect the GDG adjustment had on the results, the Bottomley figure (0.07) was used in a sensitivity analysis
Results indicate that the conclusion about cost-effectiveness changes very little using this more optimistic estimate of utility gain. The ICERs for all strategies increases relative to the base case; therefore, starting with concurrent treatment before twice daily potent corticosteroids is less likely to be cost-effective (ICER=£88,333 vs £23,250 in the base case). Similarly, the ICER for a strategy starting with TCF product increased to over £787,000 compared to starting with concurrent treatment (£174,500 in the base case).
M.3.2.5. 4-week quantity of TCF product
In the base case, hypothetical patients are assumed to use 134.0 g of TCF product during 4 weeks of treatment. Bottomley and colleagues used a much lower value for this input (92.6 g), and we explored how the results of the NCGC analysis might change if this lower estimate was used. The cost of 92.6 g of TCF product was £61.27 (compared to £94.26 in the base case). The results of this sensitivity analysis showed that the ICER for TCF product improved compared to the base case (£124,400 vs £174,545); however this is still well above the NICE cost-effectiveness threshold of £20,000 per additional QALY. Initial therapy with twice daily potent corticosteroid or concurrent vitamin D analogue and potent corticosteroid is still more likely to be considered cost-effective.
M.3.2.6. Unit costs of potent corticosteroids and vitamin D analogues
The base case assumed that the cost for each topical was based on the product and formulation with the lowest unit cost per gram/millilitre. Given that clinicians and patients may have preferences for different products or formulations, it was considered necessary to explore how variation in price of topicals, particularly potent corticosteroids and vitamin D, might affect the results. To do this, the highest cost (per gram) potent corticosteroid Synalar gel (fluocinolone acetonide) was assumed in place of Betnovate cream or ointment. The cost of Synalar gel is around four times that of Betnovate cream/ointment. In another analysis, the most costly vitamin D ointment, Curatoderm (tacalcitol), was assumed instead of Silkis (calcitriol). The cost of Curatoderm is around 2.5 times more costly than Silkis and 1.6 times more costly than Dovonex (calcipotriol) ointment. In a final sensitivity analysis, both Synalar gel and Curatoderm were used. Results in terms of incremental cost-effectiveness ratios are presented in .
Incremental cost per QALY gained under different treatment cost assumptions.
When the cost of Synalar gel is used, the ICER for starting with concurrent therapy and then moving to potent corticosteroid compared to the reverse, decreases substantially from the base case (£4,365 compared to £23,250), becoming optimal given the NICE threshold. The ICER for this strategy when only the cost of Curatoderm ointment is used and when Synalar gel and Curatoderm ointment, actually increase relative to the base case. Even with increased costs for potent corticosteroid and vitamin D, once daily TCF product is unlikely to be cost-effective compared to concurrent therapy unless the willingness to pay threshold is well over £100,000 per QALY gained.
M.3.2.7. Sensitivity analyses – Restricted comparators
The base case analysis put several conditions on the way topicals could be sequenced (see M.2.1.1). These conditions did not restrict how potent corticosteroids were fit into treatment sequences other than that they could not appear in all three lines of treatment. This included their use as part of concurrent or combined (TCF product) treatment. The GDG expressed concern that these restrictions may not fully reflect the caution they would use in prescribing trials of potent corticosteroids, in that the BNF discourages continuous use of potent corticosteroids for more than 8 weeks at a time. The GDG was also concerned that the analysis did not fully capture the safety risks associated with the continuous or intermittent use of twice daily potent steroids. In a series of sensitivity analyses, various additional restrictions were placed on the treatment sequences.
In the first scenario, it was assumed that interventions that included potent corticosteroids could not be offered consecutively. For example, once daily TCF product could not be offered after treatment with once or twice daily potent corticosteroids, nor could twice daily potent corticosteroid follow once daily potent corticosteroid. Under this assumption, starting with twice daily corticosteroid, then trying twice daily vitamin D analogue and then using both concurrently would represent the best value for NHS resources given a £20,000 per QALY threshold. Starting with concurrent treatment would only be cost-effective at thresholds of greater than £33,000 and TCF product would only be cost-effective at thresholds over £202,000.
In the second scenario, it was assumed that twice daily corticosteroid could not be prescribed as a first or second line topical therapy, but consecutive use of potent corticosteroids was permitted. Under this scenario, the optimal strategy was to start with concurrent corticosteroid and vitamin D analogue, then try twice daily vitamin D analogue alone and finally twice daily potent corticosteroid only. This had an ICER of £18,000 per QALY gained compared to once daily potent corticosteroid followed by concurrent treatment and then twice daily coal tar. Strategies including TCF product either as second or first line were not cost-effective unless the threshold was over £110,000 and £446,000, respectively.
A third scenario combined the first and second scenarios, such that twice daily potent corticosteroid could not be prescribed as first or second line treatment and no sequences could include consecutive lines of potent steroid containing strategies. Under these conditions, the same sequence as in scenario 2 is most cost-effective (Concurrent – vit D BD – PS BD). TCF product replaces twice daily steroid in that sequence only if the threshold willingness to pay is £134,000 and replaces concurrent treatment in the same sequence if the threshold is £202,000.
In a fourth and final scenario, twice daily potent corticosteroid was removed entirely and no potent steroid containing products could be prescribed consecutively. Under this assumption, the most cost-effective sequence was initial concurrent treatment followed by twice daily vitamin D alone and then twice daily coal tar. TCF product replaces twice daily coal tar in that sequence at a threshold of over £47,000 and replaces concurrent treatment at a threshold of over £489,000.
Results from all aforementioned sensitivity analyses (i.e. treatment effects, early versus late response, reduced adherence, cost of potent corticosteroids and vitamin D and so on) were reinterpreted within the context of these restricted comparator scenarios. The conclusions from each scenario presented here were insensitive to changes in the tested parameters. For example, concurrent therapy followed by twice daily vitamin D followed by twice daily potent corticosteroids was optimal across all tested parameter variation under the conditions that twice daily potent corticosteroids could not be offered as initial treatment or when steroids could not be used consecutively. Furthermore, once daily TCF product was consistently more effective but never found to have an ICER below or near to the NICE £20,000 per QALY threshold.
M.3.2.8. Downstream resource use and cost
Changes to the assumed probability of referral to secondary care and proportion offered phototherapy have no meaningful effect on the conclusions of the base case. The probability of referral to secondary care was varied downwards to 40% and upward to 80%. When referral occurred less often than in the base case, there was no change to the rank order of strategies, but the ICER for a strategy where TCF product was used first instead of concurrent treatment increased to £200,000 per additional QALY. When referral occurred more often than in the base case, there was still no change in the rank order, but the ICER for TCF product was slightly lower. If the probability of undergoing UVB phototherapy upon referral was higher than in the base case (50% vs 30%), then the ICER for TCF product compared to concurrent treatment reduced slightly, but not enough to make it cost-effective. Finally, if instead of assuming patients are treated with UVB phototherapy, it is assumed they receive outpatient day care treatment with specialist supervised topical therapies, then the ICER for concurrent therapy before potent corticosteroids alone increases to over £30,000 per QALY and the ICER for initial TCF product instead of concurrent therapy decreases to £155,000 per QALY.
If the time horizon is extended for 2 to 3 years and cumulatively more patients see a specialist and move on to UVB phototherapy, then initial treatment with concurrent vitamin D and potent corticosteroids becomes more cost-effective than starting with potent corticosteroids alone. When the time horizon is extended, TCF product becomes more cost-effective compared to concurrent treatment (ICER = £118,067 at 2 years; ICER = £90,710 at 3 years; ICER=£75,255 at 5 years; ICER=£73,541 at 10 years), but is still very unlikely to be considered cost effective given the NICE willingness to pay threshold of £20,000 per QALY gained. Visual inspection of the health state membership probabilities over a 10-year time horizon indicates that patients are no longer transitioning between health states after 8 years because they have all reached long-term management with a GP or specialist by this point. This suggests that the ICER for TCF product is unlikely to come down any further even if the model time horizon is extended beyond 10 years.