TA971 · multiple_technology_appraisal

Therapeutics for people with COVID-19

mixedFebruary 2023

Source documents

Interventions

dexamethasone (Ozurdex)
Corticosteroid · Corticosteroid · intravitreal injection
tocilizumab (RoActemra)
IL-6 receptor antagonist · humanised monoclonal antibody that inhibits the cytokine interleukin-6 (IL-6) · intravenous infusion
baricitinib (Olumiant)
Janus kinase inhibitor · Selective inhibition of Janus kinase 1 and 2 · oral
nirmatrelvir plus ritonavir (Paxlovid)
antiviral · protease inhibitor · oral
sotrovimab (Xevudy)
neutralising monoclonal antibody · binds to spike protein via ACE2 expression · intravenous
casirivimab plus imdevimab (Ronapreve)
neutralising monoclonal antibody · neutralising monoclonal antibody · intravenous
molnupiravir (Lagevrio)
antiviral · nucleoside analogue inhibiting RNA-dependent RNA polymerase · oral
remdesivir (Veklury)
antiviral · nucleotide analogue inhibiting RNA-dependent RNA polymerase · intravenous
tixagevimab plus cilgavimab (Evusheld)
neutralising monoclonal antibody · neutralising monoclonal antibody · intramuscular injection

Conditions

covid-19respiratory · mild
covid-19respiratory · severe
covid-19respiratory · acute
mild covid-19respiratory · acute
severe covid-19respiratory · acute
covid-19 (mild, non-hospitalised)respiratory · acute
covid-19 (severe, hospitalised)respiratory · acute

Comparators

NameType Established Committee preferred
standard carestandard of care
standard care at the timestandard of careYes
best supportive carebest supportive careYes

Clinical trials

TrialDesignPhasePivotal
PANORAMICRCTPhase 3
SOLIDARITYRCT
ACTT-1RCT
RECOVERYRCT
REMAP-CAPRCT
OpenSAFELYobservational
Imai et al. 2023in vitro study
EPIC-HRRCTYes
EPIC-SRRCT
PINETREERCT
COMET-ICERCTYes

Economic model

other (assessment_group)
markov

Methodological decisions (44)

comparator selection

Selection of appropriate standard care for economic model - which trial to base standard care on

Committee: Committee noted standard care arm in economic model modelled on dexamethasone arm of RECOVERY trial (UK-based, reflective of NHS practice) rather than SOLIDARITY standard care which varied across countries. RECOVERY considered more generalisable to endemic setting

ICER impact: uncertain_direction

cost assumption

Long COVID management costs: inflated from £1,013 (chronic fatigue syndrome proxy) to £2,267 per year (2021/2022 prices)

Company: Consultee provided alternative higher cost from Vos-Vromans et al. 2017

ERG: AG accepted new evidence and updated cost estimate

Committee: Agreed with updated base-case value

ICER impact: negligible

cost assumption

Administration costs for treatments: updated to include NHS England CMDU deployment costs (£410 for oral antivirals, £820 for monoclonal antibodies)

Company: Some companies disagreed with CMDU costs as based on secondary care delivery; suggested lower e-consultation/telephone triage costs (£75-£117 for nirmatrelvir plus ritonavir); suggested additional pharmacist costs (£352.49 per hour) for interaction assessment; suggested reduced costs for molnupiravir (£0 to £41)

ERG: AG included NHS England-provided CMDU deployment costs after consultation

Committee: Considered differences in administration costs in relation to net monetary benefit, noting uncertainty about future delivery models

ICER impact: increases

cost assumption

Hospitalisation costs using NHS reference codes: ordinal scales 3-7 assigned to specific DZ and XC codes with weightings for long-stay adjustments

Company: Stakeholders said approach underestimated true cost

ERG: AG agreed with changes and updated codes and adjustments

Committee: Acknowledged and agreed with AG's final approach

ICER impact: increases

indirect comparison method

Network meta-analysis approach using COVID-NMA and metaEvidence living systematic reviews. Clinical experts commented that meta-analysing trial results may not be appropriate because weighting of each trial may not consider relevance of context (e.g., different variants). Consultees noted the systematic reviews did not adhere to established methods and missed 2 key trials (SOLIDARITY and ACTT-1).

Company: Company provided alternative NMA including SOLIDARITY

ERG: Assessment group initially used NMA from living reviews; after consultation feedback, updated approach to include company-provided NMA with SOLIDARITY and scenario using ACTT-1 data for remdesivir time to discharge

Committee: Rather than use probabilistic sensitivity analysis, AG ran scenarios using mean and upper/lower confidence limits of efficacy estimates (mean efficacy, lower efficacy, higher efficacy). Committee understood limitations but considered this represented an attempt to address uncertainty. After consultation, noted heterogeneity and generalisability issues made uncertainty challenging to parameterise, so probabilistic sensitivity analysis would not capture appropriate type of uncertainty.

ICER impact: uncertain_direction

indirect comparison method

Network meta-analysis approach and appropriateness given trial context variability and evolving standard care

ERG: Network meta-analyses from publicly available sources (COVID-NMA and metaEvidence) used; AG acknowledged significant limitations including changing pandemic context, different trial designs, baseline characteristics, geographical locations, and assumption that relative treatment effects are transferable to current management

Committee: Committee recognised high levels of uncertainty with treatment effects and context-specific nature of evidence. Rather than probabilistic sensitivity analysis, used scenarios with mean, upper and lower confidence limits to characterise uncertainty

ICER impact: uncertain_direction

indirect comparison method

For mild COVID-19, committee noted that indirect comparison used pairwise analysis rather than network analysis, creating potential for bias. Multiple interventions may be required and side-by-side comparisons cautioned against.

Committee: Acknowledged limitations of pairwise indirect comparison and heterogeneity of trial outputs

ICER impact: increases

model structure

Committee questioned use of Q-COVID risk calculator in clinical practice as a model for defining high-risk groups. The committee considered Q-COVID had limited applicability because of model limitations.

Company: Not specified in this chunk

ERG: Assessment group did not use Q-COVID risk calculator; instead used PANORAMIC hospitalisation rates

Committee: Q-COVID risk calculator has limited applicability and should not be used as basis for high-risk group definition

ICER impact: negligible

model structure

AG developed decision tree for mild COVID-19 setting joining with partitioned survival model for severe COVID-19 setting

Committee: Considered model appropriate to capture important outcomes and appropriate for decision making given available evidence base

ICER impact: negligible

model structure

Low- and mean-efficacy scenarios presented for severe COVID-19 with supplemental oxygen

Committee: same assumptions as severe COVID-19 without supplemental oxygen

mortality assumption

Mortality assumptions in economic model. Consultees noted that treatment in hospital had higher mortality risk compared with standard care in low-efficacy scenario, which was considered unrealistic.

Company: Not specified in this chunk

ERG: Assessment group capped mortality rate to equal 1 for low-efficacy scenario in response to consultation feedback

Committee: Mortality rate should not exceed 1 (i.e., should not exceed untreated background mortality)

ICER impact: increases

mortality assumption

Mortality rate capping in low-efficacy scenarios to prevent unrealistic hospital mortality being higher than standard care

ERG: Initial assumption resulted in treatment in hospital having higher mortality risk than standard care

Committee: Committee accepted AG's update capping mortality rate to equal 1 for low-efficacy scenario in response to consultee feedback

ICER impact: increases

mortality assumption

Baseline hospitalisation rates: committee concluded rate for McInnes high-risk group is between 2.41-2.82%; 4.00% for people contraindicated to nirmatrelvir plus ritonavir

ERG: AG presented rates from OpenSAFELY and DISCOVER-NOW database

Committee: Concluded rate likely between 2.41-2.82% based on strength of evidence, acknowledging substantial uncertainty

ICER impact: increases

other

Whether to use probabilistic sensitivity analysis versus scenario analysis with efficacy bounds

ERG: AG ran scenarios using mean and upper/lower confidence limits rather than probabilistic sensitivity analysis due to challenges in parameterising uncertainty from heterogeneous trial populations

Committee: Committee accepted scenario analysis as appropriate attempt to address uncertainty in absence of alternative methods, acknowledging limitations

ICER impact: uncertain_direction

other

Inclusion of additional NMA analyses and trial data following draft consultation feedback

ERG: AG initially missed SOLIDARITY and ACTT-1 in systematic reviews

Committee: Committee noted AG addressed concern by providing scenarios including company-provided NMA with SOLIDARITY and scenario using ACTT-1 for remdesivir time to discharge. Also included updated COVID-19 NMA results for molnupiravir, casirivimab plus imdevimab, and tocilizumab

ICER impact: uncertain_direction

other

Committee applied an in vitro expert advisory group framework to assess neutralisation activity of monoclonal antibodies against Omicron subvariants. Casirivimab plus imdevimab and tixagevimab plus cilgavimab concluded to have insufficient neutralisation activity against most circulating variants.

Committee: Considered in vitro evidence on neutralisation activity inconsistent and highly uncertain, particularly for casirivimab plus imdevimab and tixagevimab plus cilgavimab against dominant variants

ICER impact: uncertain_direction

other

Committee considered whether reduced neutralisation in in vitro studies of sotrovimab reflected real-world clinical effectiveness. Company explained that results may depend on ACE2 expression levels in cell lines used and may underestimate sotrovimab's neutralising ability.

Company: In vitro results may underestimate sotrovimab's clinical efficacy if cell lines over-express ACE2

Committee: Concluded sotrovimab clinical effectiveness is likely reduced against BA.5 but remains uncertain against BQ.1 and BQ.1.1; in vitro evidence was ambiguous

ICER impact: uncertain_direction

other

PANORAMIC published results showed no significant difference on hospitalisation or mortality versus standard care, only secondary endpoint of time to recovery was significant. Committee concluded clinical efficacy in hospitalisation and mortality is uncertain.

Committee: Could not be certain of molnupiravir's clinical efficacy in hospitalisation and mortality when potential benefit is minimal

ICER impact: decreases

other

Committee noted statistically significant reduced risk of hospitalisation from PINETREE but lacked evidence of survival benefit. Considered all efficacy estimates due to uncertainty.

Committee: Considered all efficacy estimates because of uncertainty

ICER impact: uncertain_direction

other

COMET-ICE showed significant reduction in hospitalisation or death but conducted before Delta wave. OpenSAFELY data during BA.2/BA.5 showed no evidence of difference versus nirmatrelvir. Committee preferred low-efficacy scenario due to reduced neutralisation against BQ variants.

Committee: Preferred low-efficacy and scenario between mean and low efficacy for sotrovimab rather than mean-efficacy scenario

ICER impact: decreases

other

Committee noted tocilizumab acts on virus complications rather than neutralising virus itself, making mechanism more robust to variant changes compared with neutralising monoclonal antibodies.

Committee: More confident in tocilizumab mean-efficacy results because of robust mechanism of action and stronger clinical trial evidence base

ICER impact: increases

other

AG assumed 100% of severe COVID-19 patients and 10% of mild COVID-19 patients would have long COVID. Consultees noted proportions should be reduced for severe setting and increased for mild setting.

Committee: AG considered assumptions conservative and appropriate despite consultee disagreement; alternative evidence was not available at time of second meeting

ICER impact: uncertain_direction

other

Committee noted AG initially used time to discharge data from ACTT-1 for remdesivir but this was collected early in pandemic with generalisability concerns. AG created scenarios removing treatment effects on time to discharge and clinical improvement at 28 days.

Committee: Concluded reasonable to remove treatment effects on time to discharge due to uncertainty about endemic setting benefits and multiple factors affecting discharge timing (e.g. testing requirements)

ICER impact: decreases

population generalisability

Clinical evidence largely from studies before dominant Omicron variants; committee noted evolving SARS-CoV-2 virus and changing epidemiology (BQ.1, XBB.1.5, CH.1.1 subvariants now dominant) affects generalisability of trial evidence to current endemic context

Committee: Consider relevance of clinical data to current endemic context despite ongoing rapid changes in variants and disease characteristics

population generalisability

Marketing authorisations based on populations with different definitions of high-risk status; some trials included people with at least 1 risk factor whereas others had specific age requirements; heterogeneous population representation

Committee: Clearly define high risk to determine treatment eligibility rather than rely on marketing authorisation definitions

population generalisability

The committee questioned whether PANORAMIC trial population (including people aged over 50 years with no comorbidities) represented the high-risk population eligible for treatment. PANORAMIC had a broader risk definition than the McInnes high-risk criteria used in NHS interim commissioning policy.

Company: Not specified in this chunk

ERG: Assessment group acknowledged that PANORAMIC included people with lower risk of severe COVID-19 compared with McInnes definition

Committee: The committee concluded that McInnes report definition of high risk was the most robust definition of people at high risk for progression to severe COVID-19, and this did not include age as an independent risk factor.

ICER impact: uncertain_direction

population generalisability

Impact of rapidly evolving SARS-CoV-2 variants on generalisability of trial evidence. Each trial conducted at different time in pandemic with different dominant variants (earlier trials pre-Omicron, later trials with Omicron sublineages BA.5, BQ.1, XBB.1.5, CH.1.1). Standard care evolved with better understanding of disease and use of dexamethasone.

Company: Not specified in this chunk

ERG: Assessment group used network meta-analyses from living systematic reviews and assumed relative treatment effects are transferable to current clinical management, but acknowledged significant context-specific limitations

Committee: Committee recognised high levels of uncertainty with each treatment effect and context-specific nature of evidence. Most appropriate approach is to consider how relevant clinical data are to current endemic context of disease at time of evaluation.

ICER impact: uncertain_direction

population generalisability

Applicability of trial evidence from pandemic setting to current endemic setting with different population immunity, viral pathogenicity, and supportive care

Committee: Committee acknowledged main generalisability concerns including changes in population immunity through vaccination and natural immunity, virus pathogenicity, and effectiveness of supportive care. Considered absolute changes in modelling where possible but relative risks from trials would lack generalisability

ICER impact: decreases

population generalisability

EPIC-HR evidence from unvaccinated population in earlier pandemic wave. EPIC-SR preliminary results in vaccinated high-risk group showed non-significant reduction in hospitalisation. Committee noted these generalisability concerns.

Committee: Considered range between mean- and lower-efficacy estimates more suited to endemic setting despite limitations

ICER impact: decreases

population generalisability

SOLIDARITY and updated NMA data for mortality benefit in severe COVID-19. Committee noted confidence intervals reflect population characteristics and standard care practices from earlier pandemic. Standard care has significantly changed since SOLIDARITY was conducted.

Committee: Interpreted evidence with caution; standard of care for severe COVID-19 has substantially changed since SOLIDARITY. Considered threshold analysis using mortality rate ratios 0.85 to 1.00, expecting ratio to tend towards 1.00 due to generalisability concerns.

ICER impact: decreases

population generalisability

Multiple potential sources for hospitalisation rates. AG used 0.77% from PANORAMIC for base case but consultees highlighted PANORAMIC excluded higher-risk eligible patients. Other rates: OpenSAFELY 2.41%, 1.37% without contraindications, DISCOVER-NOW 2.82%, advanced renal disease 4.15%-4.4%, immunocompromised 15.9%.

Committee: Concluded hospitalisation rate for McInnes high-risk group likely falls between 2.41%-2.82% based on OpenSAFELY and DISCOVER-NOW; for contraindicated patients approximately 4% using advanced renal disease as proxy

ICER impact: decreases

population generalisability

PANORAMIC trial may have underrepresented highest-risk patients; committee preferred McInnes high-risk definition over PANORAMIC population

ERG: AG noted PANORAMIC would have excluded higher-risk patients eligible under NHS interim clinical commissioning policies

Committee: Preferred hospitalisation rates between 2.41-2.82% (McInnes definition) over PANORAMIC's 0.77%; acknowledged 0.77% likely underestimation

ICER impact: uncertain_direction

population generalisability

Committee evaluated high-risk definition using McInnes report for nirmatrelvir plus ritonavir recommendations, narrower than marketing authorisation

Committee: McInnes defined high-risk group for cost-effectiveness threshold

subgroup definition

Debate over whether to define separate 'high risk' and 'highest risk' groups, or maintain a single high-risk group. The committee considered splitting high-risk groups based on immunosuppression status (e.g., people having rituximab) versus other systemic therapies, but concluded insufficient evidence was available at subgroup level.

Company: Not specified in this chunk

ERG: Assessment group assumed general population survival with starting age 56.6 years and same hospitalisation rate as PANORAMIC; no individual high-risk subgroups modelled based on specific baseline characteristics

Committee: Single definition of high risk should be used. Evidence at subgroup level is limited and too uncertain to parameterise the model.

ICER impact: uncertain_direction

surrogate endpoint validity

Use of in vitro neutralisation assays to infer clinical effectiveness of treatments against evolving variants

Company: Company provided observational data on remdesivir treatment effect during draft consultation and explained sotrovimab's effectiveness depends on ACE2 expression levels which may be over-expressed in cell lines, potentially underestimating neutralising ability

Committee: Committee commissioned in vitro expert advisory group to develop decision framework linking in vitro neutralisation data to clinical outcomes. Framework accepted by clinical experts. For tocilizumab, clinical effectiveness not variant-specific due to mechanism of action. For antivirals, no in vitro evidence showing reduced efficacy across variants tested. For monoclonal antibodies, they concluded casirivimab plus imdevimab and tixagevimab plus cilgavimab unlikely to retain sufficient neutralisation activity against most circulating variants

ICER impact: decreases

surrogate endpoint validity

Clinical experts noted hospitalisation and mortality rates becoming less relevant for COVID-19 treatments due to changing landscape

Committee: current model captures relevant outcomes; some benefits fall outside NICE reference case

treatment effect duration

Cost-effectiveness estimates highly dependent on treatment efficacy and hospitalisation and mortality rates; hospitalisation and mortality rates lower with Omicron variants than earlier pandemic variants, which increases cost-effectiveness estimates

ICER impact: increases

treatment effect duration

AG initially used 108.6 weeks for long COVID duration. Consultees highlighted this was underestimated. AG updated model estimating 30% with symptoms at 2 years, 10% at 5 years, 3% at 10 years.

Committee: AG considered its assumptions conservative and appropriate despite consultee feedback because alternative evidence was unavailable

ICER impact: decreases

treatment effect duration

Removal of treatment effects on time to discharge and clinical improvement at 28 days in scenarios; HR set to 1 for these outcomes

ERG: AG included scenarios removing these treatment effects to account for uncertainties

Committee: Concluded it was reasonable to remove these treatment effects, considering heterogeneous population and changed clinical context in endemic setting

ICER impact: decreases

treatment effect duration

Remdesivir: threshold analysis of mortality rate ratios between 0.85 and 1.00 conducted

Committee: insufficient evidence to show meaningful mortality benefit

treatment effect waning

Whether relative treatment effects from pandemic setting trials remain applicable to endemic setting with changed vaccination, natural immunity, pathogenicity, and supportive care

Committee: Committee concluded mean-efficacy scenarios likely reflect ceiling efficacy; relative treatment effects would lack generalisability due to changes in best supportive care and vaccination rates. Effects would tend towards hazard ratio of 1 in endemic setting

ICER impact: decreases

utility source

Use of UK age- and sex-adjusted EQ-5D-3L utility values for baseline estimates; no additional utility decrements for mild COVID-19 without long COVID

ERG: AG used age- and sex-adjusted UK general population utility estimates

Committee: Agreed with AG's assumption; acknowledged minor impact on ICERs

ICER impact: negligible

utility value choice

Use of utility decrements from cost-effectiveness analysis of remdesivir (originally from Clostridioides difficile and influenza populations) applied across ordinal scales 3-5 for severe COVID-19

Company: Not stated

ERG: AG used utility decrements from non-COVID-19 population

Committee: Agreed with AG; did not find alternative COVID-19-specific decrements from stakeholder-suggested systematic reviews

ICER impact: uncertain_direction

utility value choice

Use of post-discharge long COVID utility decrements from Evans et al. 2022, same decrement regardless of ordinal scale status at hospital admission

ERG: AG used uniform post-discharge utility decrement

Committee: Agreed with AG's rationale; noted minimal impact on ICERs

ICER impact: negligible

Evidence gaps

surrogate not validatedNo discussion of surrogate endpoints, but context of rapidly evolving variants creates uncertainty about whether treatment effects on hospitalisation/mortality from past trials transfer to current Omicron-dominant setting
short follow upTrial evidence from different periods of pandemic; most recent trials conducted with Omicron variants (BA.5, BQ.1), but newer subvariants (XBB.1.5, CH.1.1) emerging. Unclear how rapidly treatment effects may wane as variants continue to evolve.
no uk dataCommittee noted that evidence at subgroup level (e.g., different immunosuppressed populations) is limited. McInnes report definition has outcomes data collected from UK vaccinated population infected with Omicron variants, but separate subgroup evidence is lacking.
single arm evidence onlyMeta-analyses from trials conducted at different times with evolving standard care. Clinical experts stated meta-analysing trial results may not be appropriate due to context-specific differences.
surrogate not validatedIn vitro neutralisation assays may not reliably predict clinical effectiveness; interpretation uncertain without additional clinical evidence such as in vivo or pharmacokinetic data
short follow upNeutralising monoclonal antibodies demonstrate variant-specific effects; clinical effectiveness likely to change as virus evolves
no direct comparisonRelative treatment effects for mild COVID-19 based on pairwise indirect comparison rather than network meta-analysis
short follow upIn vitro evidence on sotrovimab neutralisation against BQ.1 and BQ.1.1 subvariants remains uncertain; no clinical data available for these variants at time of evaluation
surrogate not validatedCommittee noted that reduced in vitro neutralisation does not necessarily equate to reduced clinical efficacy; unclear how much reduced neutralising effect impacts clinical efficacy
short follow upEPIC-SR preliminary results with early termination of enrolment; published only on company website rather than peer-reviewed journal
single arm evidence onlyCOMET-ICE remdesivir evidence lacks comparator control in some analyses
no uk dataLimited UK-specific evidence on treatment effects in endemic COVID-19 setting; reliance on international trials from earlier pandemic waves

Commercial arrangement

none · confidential

Special considerations

Innovation acknowledged Equality issues raised