Preprints.org logo
Preprint
Article

This version is not peer-reviewed.

Completeness Before Certainty: Evidence Paradoxes in the First EU Joint Clinical Assessments

Submitted:

28 October 2026

Posted:

29 October 2026

You are already at the latest version

Abstract
The EU Health Technology Assessment Regulation makes the joint clinical assessment dossier the foundation of a single European appraisal of the relative effectiveness and safety of new medicines, and Article 9 conditions that foundation on the completeness of the evidence a developer submits. Drawing only on the first publicly observable outputs of the new system — two completeness reviews and one completed assessment — this article asks how the European Commission and the HTA Coordination Group are giving operational meaning to “completeness”. It argues that an exacting interpretation is emerging, closer to the logic of systematic review than of regulatory submission, but that the evidentiary base remains too thin to call it a settled doctrine. Two structural tensions follow: a certainty paradox, in which uncertainty is catalogued exhaustively yet never converted into a graded judgment of certainty; and an evidence-generation paradox, in which developers of orphan and highly innovative therapies may face analytical demands whose outcome is largely foreseeable. The article situates these tensions within the principle of proportionality, the duty to give reasons and the limits of judicial review, and sets out a research agenda for the Regulation’s scheduled 2028 evaluation.
Keywords: 
health technology assessment
;  
regulation under uncertainty
;  
proportionality
Subject: 
Public Health and Healthcare  -   Health Policy and Services

Introduction

Health technology assessment (HTA) is the process by which health systems evaluate the added value of a medicine, medical device or other health technology relative to the options already in use, in order to inform decisions about whether, and on what terms, to fund it. It must be distinguished from the regulatory approval that precedes it. Before a medicine reaches patients in Europe, it must obtain a marketing authorisation, as a rule from the European Medicines Agency, which asks whether the product is safe, efficacious and of adequate quality in absolute terms. HTA asks a different, comparative question: not “does this work?” but “does it work better than what we already use, and by how much?” [1] The answer feeds national decisions on pricing and reimbursement that determine real-world access. HTA therefore sits at the interface of science, economics and law — a scientific appraisal of evidence that carries direct distributive consequences and, increasingly, legal ones.
Historically, HTA has been a national competence, and for the questions that bear most directly on national budgets it remains one. Each Member State has developed its own methods, timelines and evidentiary expectations, so that the same medicine could be assessed many times over, to different standards, yielding divergent verdicts on the same underlying trials. For two decades this duplication was tempered only by voluntary cooperation, latterly through the EUnetHTA joint actions, which produced shared assessments that Member States were free to use or to ignore. [2] The inefficiency for developers, and the inconsistency for patients, supplied the political impetus for a binding instrument.
That instrument is Regulation (EU) 2021/2282, the Health Technology Assessment Regulation (HTAR), [3] which for the first time makes part of the clinical appraisal mandatory and common across the Union. Its central mechanism is the joint clinical assessment (JCA): a single, Union-level evaluation of a medicine’s relative clinical effectiveness and relative safety, conducted once by assessors drawn from national bodies and coordinated through a Member State Coordination Group, in place of the parallel national clinical assessments that preceded it. The JCA is deliberately confined to the clinical core. It does not set price, it does not decide reimbursement, and it does not pronounce on a technology’s overall “value”; those choices remain with each Member State, which layers its own economic and social appraisal onto the common clinical foundation. The JCA is thus designed as a shared factual base on which national decisions are built, not as a decision in itself. This division of labour — common clinical facts, national value judgments — is what makes the integrity of the JCA dossier so consequential, and it is where the present inquiry begins.
This is the most significant reform of European HTA cooperation since it began. The HTAR became applicable to the first categories of medicinal products — new active substances for the treatment of cancer, and advanced therapy medicinal products — in January 2025, with orphan medicinal products following in 2028 and all remaining products by 2030.
The fulcrum of this design is Article 9. The credibility of a single European assessment depends on the quality and completeness of what the developer submits, and Article 9 translates that dependence into a legal requirement: the dossier must be complete, the evidence appropriately analysed, the presentation transparent and the underlying documentation supplied. [4] Yet until recently the practical meaning of “completeness” was largely undefined. The Regulation states the principle; the implementing measures and methodological guidance supply procedural detail; [5] but the operational content of concepts such as “complete and up-to-date information”, “appropriate methods” and “transparent presentation” could only be discovered through implementation.
The first outputs of the system now allow that discovery to begin. Two of the earliest publicly observable documents are notices issued under Article 10(5), recording the information the Commission found missing from a dossier after a second request: one concerning an autologous tumour-infiltrating lymphocyte (TIL) therapy, [6] the other concerning catequentinib. [7] A third document is a completed assessment at the hardest edge of the framework — an orphan medicine for a rare paediatric cancer, tovorafenib. [8] Together these documents are something close to a natural experiment: they show the framework behaving under stress, and they let us read its unwritten assumptions.
This article makes three claims. First, an exacting interpretation of Article 9 completeness is emerging, one that treats the dossier as an auditable evidence-synthesis document rather than a curated submission of favourable results. Second, that inference must be disciplined: two second-request notices from a single regulator in the opening months of a new regime can ground a working hypothesis, not a doctrine, and the completed assessment is a third observation of a different kind. [9] Third, two structural tensions follow from the emerging standard — a certainty paradox and an evidence-generation paradox — that engage the principle of proportionality and the governance of regulation under uncertainty. [10] Section II sets out the Article 9 architecture; Section III introduces the three cases; Section IV states the emerging standard and its evidentiary limits; Sections V and VI develop the two paradoxes; Section VII turns to governance and legal implications; Section VIII addresses the 2028 evaluation.

II. The Article 9 Completeness Standard in the HTAR Architecture

1. The dossier-based design and the Article 10 gateway
The HTAR establishes what appears to be a clean division of labour. The developer generates and submits evidence; the assessor and co-assessor evaluate it. Before a joint clinical assessment proceeds, the Commission checks that the submitted dossier meets the Article 9 requirements — a completeness review that operates as the gateway to assessment. [11] Where the dossier is found wanting, the Commission may issue a first and then a second request for the missing information. [12] If, after the second request, the dossier still fails to meet the Article 9(2), (3) and (4) requirements, the Coordination Group is to discontinue the joint clinical assessment. [13]
The reason the dossier carries so much weight is the ambition of the reform it serves. The HTAR seeks to replace the duplication of national clinical assessments with a single Union-level appraisal of the clinical core — relative effectiveness and relative safety — while leaving the appraisal of value, price and reimbursement to the Member States. That settlement makes the joint assessment the shared factual basis on which twenty-seven national systems are expected to build. If the shared basis is incomplete or opaque, the harmonisation rationale fails at its foundation, and the national divergence the Regulation set out to reduce simply migrates downstream. Article 9 is therefore not a clerical checklist but the hinge on which the credibility of the whole construction turns.
Two features of the gateway matter for what follows. The notices analysed in this article are second-request documents: they record deficiencies that survived an earlier round of correction, and they sit one step short of discontinuation. They are therefore not routine administrative queries but considered statements of what the Commission regards as indispensable. At the same time, they are not assessments of the medicine; they are judgments about the dossier. The distinction is easy to lose but analytically essential, because a deficiency notice tells us what completeness requires without telling us anything about whether the medicine works. It also follows that these documents are drawn from the difficult tail of the caseload — a point that bears directly, in Section IV, on how much can responsibly be inferred from them.
2. Completeness as a cumulative, multidimensional condition
Article 9 does not define completeness as a single attribute. Article 9(2) requires complete and up-to-date information to assess the parameters in the assessment scope. Article 9(3) adds four cumulative conditions: the submitted evidence must be complete with regard to the available studies and data (sub-paragraph (a)); the data must be analysed using appropriate methods to answer all research questions (sub-paragraph (b)); the presentation must be well structured and transparent so that assessment is possible within tight timeframes (sub-paragraph (c)); and the dossier must include the underlying documentation (sub-paragraph (d)). Article 9(4) then requires the information set out in Annex I.
Read together, these provisions make completeness multidimensional. It is not enough that relevant evidence exists, or even that favourable results are presented. The developer must demonstrate that the evidence was comprehensively identified, appropriately analysed, transparently presented and fully documented. Because the conditions are cumulative, a dossier may fail on any one of them. That structure — a chain of necessary conditions spanning the whole evidence process — is the background against which the early cases should be read.
3. Methodological guidance as the unwritten layer
The statutory text does not, by itself, fix what any of these conditions concretely require. That work is done by the methodological guidance the Coordination Group is empowered to develop, [14] which specifies how searches should be conducted, how indirect comparisons should be approached and how results should be presented. The early cases make the importance of this layer plain: the catequentinib notice faults the developer for departing from guidance on study selection, and the tovorafenib assessment is structured throughout by reference to it. Two implications follow. First, the operational meaning of completeness lives as much in soft guidance as in the Regulation, [15] which makes the standard more malleable — and more capable of adaptation — than the bare text suggests. Second, the guidance layer is precisely where the tensions identified in this article could be addressed: a graded framework for expressing certainty, or an orphan-specific methodological adaptation, would most naturally be introduced there rather than by amending the Regulation.

III. Three Early Cases as a Natural Experiment

1. The TIL therapy: reproducibility and the analytical record
The TIL-therapy completeness review is dominated by demands going to the integrity of the analytical record. The notice asks for relative-effect measures rather than bare event proportions for several populations; for descriptions and justifications of the analytical methods used; for the elements needed to judge the certainty of an unanchored indirect comparison; for risk-of-bias documentation tied to each signalling question of the assessment tool; and, strikingly, for the program code underlying the indirect comparison. [16] The recurring theme is reproducibility: complex analyses are evaluated not only on their results but on whether an independent assessor could follow, verify and reproduce them.
Several of the requests are revealing precisely because they are so granular. A demand for the program code of an indirect comparison treats the analysis as something to be re-executed, not merely read. A demand for material tied to each signalling question of a named risk-of-bias instrument treats bias assessment as a documented chain of reasoning rather than a summary verdict. And a demand for documentation supporting a statement that a product manufactured at one site was not comparable to the same product manufactured at another shows the Commission probing the factual claims on which the developer’s own analysis rested. The cumulative message is that, for a complex evidence package, completeness means auditability: the dossier must enable the assessor to reconstruct the developer’s reasoning end to end.
2. Catequentinib: evidence identification and scope
The catequentinib review, addressed to an updated dossier for two sarcoma indications, has a different centre of gravity. Its demands concentrate on the identification of evidence and the definition of scope: a transparent, comprehensive search for all available evidence; reasons for not assessing the feasibility of indirect treatment comparisons; a justification for departing from methodological guidance by not searching for non-randomised studies; [17] consistent reporting of searches; lists of excluded studies with reasons; and systematic searches of national HTA-agency websites and patient registries. [18] For one of the two indications, the Commission records that no data or analyses on relative effectiveness or safety were provided for any of the populations in scope — a gap that goes not to documentation but to the existence of comparative evidence at all.
Two aspects deserve emphasis. First, the notice does not merely ask for more studies; it asks the developer to justify the architecture of its evidence search — why certain study designs were excluded, why feasibility assessments were not performed, why guidance was not followed. Completeness here is procedural and justificatory: the developer must account for the choices that shaped the evidence base, not only report its contents. Second, the indication for which no comparative data were supplied illustrates the limiting case of incompleteness, where the deficiency is not that the evidence was poorly presented but that the comparative question was, in effect, left unanswered. The two indications within a single product thus span the spectrum from a documentation gap to a substantive evidentiary void.
3. Tovorafenib: a completed assessment at the orphan edge
The tovorafenib assessment differs in kind from the two notices: the dossier cleared the gateway and was assessed. It is nonetheless instructive, because it shows what happens when the framework is applied to an orphan medicine supported principally by a single-arm study with no common comparator. Of the populations originally in scope, only one could be assessed comparatively, on a surrogate endpoint and at a very small effective sample size; the remainder could not be assessed at all. [19] The assessors did not passively accept the dossier: they recomputed missing relative effects and reproduced the weighting of the indirect comparison — rigorous verification that, as Section VII discusses, also blurs the assumed boundary between assessment and evidence generation.
Two characteristics of the report carry beyond the single product. The first is the asymmetry between the exhaustive treatment of uncertainty and the absence of a corresponding statement of certainty, examined in Section V. The second is the collapse of scope itself: a framework built around the comparison of a new medicine against a relevant alternative encountered evidence that could not, for most populations, sustain any comparison at all. That collapse is not an artefact of a careless dossier; it is what the comparative-effectiveness paradigm produces when it meets orphan evidence. The case therefore functions less as a verdict on one medicine than as a stress test of the framework at its hardest edge, and it is in that diagnostic spirit that it is used here.

IV. An Emerging Completeness Standard — and the Limits of Inference

1. What the two completeness reviews share
Beneath their different emphases, the two notices share a conception of completeness. In both, completeness is treated as a demonstrable property of the entire evidence process — from the identification of studies, through their selection and the justification of exclusions, to the methods of analysis and the inputs needed to assess certainty — rather than as a property of the results alone. The TIL-therapy review polices the downstream end of that process (analysis, presentation, documentation); the catequentinib review polices the upstream end (search, selection, scope). The Annex I requirement in Article 9(4) functions largely as a catch-all that restates the others rather than adding a distinct category. The common move is to ask the developer to make the whole evidentiary chain auditable.
This is a demanding reading. It aligns the JCA dossier more closely with the conventions of a systematic review — pre-specified searches, transparent selection, documented exclusions, reproducible synthesis — than with the conventions of a regulatory submission designed to demonstrate benefit. If it hardens, it will reshape how developers prepare dossiers, because it makes the process of evidence handling, and not only its product, a condition of admissibility.
The shift is best understood as one of paradigm. A regulatory submission is advocacy within rules: the developer marshals the available evidence to support a claim, and the assessor tests that claim. A systematic review is a method for establishing what the evidence collectively shows, judged by the transparency and reproducibility of the process that produced it. The early cases suggest the Commission is reading Article 9 in the second register. The consequence is that a dossier can be complete in the submission sense — every favourable study included, every result reported — yet incomplete in the synthesis sense, because the search was not systematic, the exclusions were not justified, or the analysis could not be reproduced. That gap between the two senses of completeness is, on the present evidence, where dossiers are failing. The framework is designed to prevent the evidentiary picture from depending on which studies happen to be retrieved, published, or selected for inclusion.
2. Why two cases cannot make a doctrine
The temptation is to call this an Article 9 doctrine. That label should be resisted, for reasons of method rather than modesty. Two second-request notices, issued by one institution in the first months of a new regime, are a very small and possibly unrepresentative sample. Second-request notices are, by construction, drawn from the troubled tail of the distribution; they tell us what the Commission demanded when a dossier was already in difficulty, not how completeness is assessed across the run of cases. The parallel between two such documents may reflect a stable interpretation, or it may reflect two dossiers that happened to share similar weaknesses. Time will revealed the interpretation.
The tovorafenib assessment does not cure this problem, because it is a different kind of observation: a completed assessment rather than a discontinued or deficient dossier. It can corroborate claims about how the framework treats uncertainty — as the next section argues — but it cannot add a third instance to a pattern of completeness findings, because the dossier was found complete enough to assess. The honest conclusion is that an exacting completeness standard is a well-founded hypothesis about the framework’s direction, to be tested as further cases accumulate. The Regulation’s own evaluation timetable, discussed in Section VIII, is the natural moment for that test.

V. The Certainty Paradox

1. Uncertainty catalogued, certainty ungraded
The assessment reporting template and the methodological guidance adopted by the HTA Coordination Group call for the certainty of the relative effects to be characterised. [20] The tovorafenib assessment documents uncertainty in exhaustive detail — residual confounding, comparability of populations, small effective samples, an uncertain proportional-hazards assumption, concerns about the causal interpretation of survival curves — yet it stops short of converting that catalogue into a graded conclusion of the kind familiar from structured certainty frameworks: high, moderate, low or very low. [21] The reader learns, in great detail, why the evidence is uncertain; the reader does not learn how that uncertainty resolves into a degree of confidence.
The contrast with established practice sharpens the point. Structured approaches to grading evidence — of which the GRADE system is the best known — begin from a starting level of certainty fixed by study design and then move that level up or down through transparent, named domains, arriving at an explicit final category. [22] The tovorafenib report performs much of the diagnostic work of such a system, identifying the domains that erode confidence, but it never completes the final, decisive step of assigning a level. The machinery of certainty assessment is present; its output is withheld. Whether by design or by the absence of an agreed framework, the effect is the same: the document tells the reader why to be uncertain, but not how uncertain to be.
It might be objected that this reticence is deliberate. The Regulation confines the joint assessment to the clinical core, providing that the assessment report shall not contain any value judgment or conclusion on the overall clinical added value of the technology and shall be limited to a description of the scientific analysis of its relative effects; [23] a graded verdict, on this view, would be precisely the kind of judgment the framework reserves to the Member States. The objection does not hold, and seeing why matters for the argument. Grading the certainty of a relative effect is not a value judgment on added value; it is part of the description of the scientific analysis that the provision expressly demands. To state that a hazard ratio is estimated with low certainty — because of residual confounding and a fragile proportional-hazards assumption — is a statement about the evidence, not about the worth of the medicine or the price a health system should pay for it. Far from straying beyond the JCA’s remit, an explicit grading would discharge that remit more completely: the reporting template and the methodological guidance already require the certainty of the relative effects to be characterised, and a graded conclusion is simply that characterisation carried to its natural end-point. The gap is therefore not a boundary the framework has chosen to respect; it is a step the framework itself calls for, but which the early practice has not yet taken.
The two completeness reviews illuminate the same gap from the other side. Both repeatedly demand the inputs required to assess certainty — in the TIL-therapy review, the elements needed to judge the certainty of the external comparison; in catequentinib, the information needed to assess the certainty of relative effects for a defined population. The system, in other words, insists on collecting the raw materials of a certainty judgment while declining to deliver the judgment itself. This is the certainty paradox: certainty is treated as something to be supported, not stated.
2. Displacement onto the national reader
The consequence is operational, not merely presentational. A national appraisal body cannot act on an inventory of limitations; it must decide how much confidence to place in each estimate. Where the joint assessment stops at uncertainty and leaves the conversion to certainty unstated, that decisive step is displaced onto twenty-seven national readers. The same evidence may then support different conclusions in different jurisdictions for interpretive rather than scientific reasons — which is precisely the divergence the joint assessment was meant to reduce. A framework built to harmonise the clinical core of appraisal risks re-importing heterogeneity at the very point where confidence is converted into decision.

VI. The Evidence-Generation Paradox

1. When the conclusion is foreseeable
For some classes of medicine, the destination is visible from the outset. Where an orphan therapy rests on a single-arm study in a small population with no common comparator, the structural limitations of the evidence are not failures of diligence; they are features of the setting. The catequentinib indication for which no comparative data were provided at all, and the collapse of the tovorafenib scope from many populations to one assessable on a surrogate, are different expressions of the same condition: comparative effectiveness is, in these settings, very hard to demonstrate and sometimes impossible to manufacture without withholding a promising therapy from desperately ill patients.
This yields the evidence-generation paradox. A developer may be asked to invest very substantial analytical effort — systematic searches, feasibility assessments, indirect comparisons, sensitivity analyses, reproducible code — to produce a public document whose dominant message, that the evidence is highly uncertain, was foreseeable before the work began. The paradox is best stated from the developer’s perspective and as a hypothesis about incentives, not as a finding: the point is not that limitations should go unreported, for transparency is essential, but that the balance between analytical burden and informational gain may not always be proportionate.
2. Proportionality and innovation incentives
Stated this way, the paradox is a proportionality concern in the ordinary regulatory sense. Proportionality asks whether a measure is suitable and necessary to its aim and whether its burdens are excessive in relation to its benefits. [24] Applied here, the suitability limb is largely satisfied: rigorous evidence requirements plainly serve the aim of a credible common assessment. The strain appears at the necessity and stricto sensu limbs. A completeness standard calibrated to the comparative-effectiveness paradigm may be suitable for the common-disease evidence it was designed around, yet impose burdens that are not necessary — because no amount of additional analysis can supply a comparator that does not exist — and that are excessive in relation to the marginal information they yield where the conclusion of high uncertainty is effectively predetermined by the evidentiary setting.
The medium-term worry is systemic rather than individual. If rare-disease developers anticipate exacting demands and foreseeable conclusions of high uncertainty, the framework may subtly discount an entire class of medicines for being what they unavoidably are, with consequences for the incentives to invest where evidence will always be thin. This is not an argument for lowering evidentiary standards; weak evidence should be identified as such. It is an argument that the framework should distinguish, openly, between uncertainty that diligence could have reduced and uncertainty that the disease itself imposes — a distinction the early cases suggest is not yet being drawn.

VIII. Implications for the 2028 Evaluation

The Regulation contains its own mechanism for testing the hypotheses advanced here. Article 31 requires the Commission to evaluate the application of the Regulation and report, providing a scheduled occasion to ask whether the early patterns are transitional artefacts of a system finding its feet or structural features that will harden with use. [30] Three questions should be on that agenda. Does the exacting completeness standard recur across a representative run of cases, including dossiers that clear the gateway without difficulty? Is the certainty paradox addressed by adopting an explicit, graded framework for expressing certainty, so that the conversion from uncertainty to confidence is performed once at Union level rather than twenty-seven times nationally? And does the evidence-generation paradox warrant an orphan-specific methodological adaptation, so that rarity-driven uncertainty is not appraised as though it were avoidable methodological weakness?
Each question is answerable with the kind of evidence the system itself generates. Completeness findings can be coded and counted; the presence or absence of graded certainty conclusions can be audited across reports; the proportion of orphan assessments ending in unassessable scope can be tracked. The argument of this article is, in that sense, a set of falsifiable propositions rather than a verdict, and the 2028 evaluation is where they should be tested.

IX. Conclusion

The European Union’s first joint clinical assessments reveal a system that is, so far, more articulate about what must be submitted than about how much confidence to place in what it concludes. An exacting interpretation of Article 9 completeness is emerging, demanding that the whole evidentiary process be made auditable; but two cases cannot make a doctrine, and the claim is best held as a hypothesis for the Regulation’s own evaluation to test. From that emerging standard two tensions follow. The certainty paradox leaves the decisive conversion from uncertainty to confidence unstated, displacing it onto national bodies and risking the very divergence the framework was meant to reduce. The evidence-generation paradox asks developers of orphan and innovative therapies to meet demands whose outcome may be foreseeable, raising a proportionality question with consequences for innovation incentives.
These are not defects to be eliminated so much as tensions to be governed. Completeness before certainty may be a defensible sequence for a young system: better to insist first on a transparent evidentiary record and to mature the expression of certainty later. But the sequence has costs, and they fall most heavily on the medicines whose evidence is hardest to generate. Whether the patterns traced here are the growing pains of a new framework or the early outline of its permanent shape is, for now, genuinely open — and that is precisely why it is worth watching with discipline rather than deciding in advance.

Competing interests

The author declares none.

References

  1. Regulation (EU) 2021/2282 of the European Parliament and of the Council of 15 December 2021 on health technology assessment and amending Directive 2011/24/EU [2021] OJ L 458/1 (‘HTAR’), art 2.
  2. On the prior voluntary cooperation and the limits that motivated a binding instrument, see HTAR (n 1) recitals 1–4.
  3. HTAR (n 1) 1.
  4. HTAR (n 1), Art. 9(2)–(4).
  5. Commission Implementing Regulation (EU) 2024/1381 of 23 May 2024 laying down, pursuant to Regulation (EU) 2021/2282 on health technology assessment, procedural rules for the interaction during, exchange of information on, and participation in, the preparation and update of joint clinical assessments of medicinal products for human use at Union level, as well as templates for those joint clinical assessments [2024] OJ L, 24.5.2024 (‘Implementing Regulation’).
  6. European Commission, ‘Information, data, analyses and other evidence missing from the dossier of the joint clinical assessment’ (Case JCA-MP-2025-02, 13 February 2026) (‘TIL-therapy completeness review’). The notice names the medicinal product by a proprietary name; as no international non-proprietary name has been assigned, and consistent with the naming convention used throughout, the product is referred to here as an autologous tumour-infiltrating lymphocyte (TIL) advanced therapy.
  7. European Commission, ‘Information, data, analyses and other evidence missing from the updated dossier of the joint clinical assessment of the medicinal product Catequentinib’ (Case JCA-MP-2024-03, 8 May 2026) (‘catequentinib completeness review’).
  8. Member State Coordination Group on Health Technology Assessment, Joint Clinical Assessment Report of Tovorafenib (Case JCA-MP-2024-06, Version 1.0, 2026), endorsed by the HTA Coordination Group on 30 April 2026, procedural review by the Commission concluded 19 May 2026, published 9 June 2026.
  9. A methodological note is owed: the two completeness reviews are second-request notices under Art. 10(5) and are, by their nature, drawn from the more difficult cases, while the tovorafenib report is a single completed assessment. They are treated here as early signals of an interpretive direction rather than as proof of a settled practice; the point is developed in Section IV.2.
  10. On regulation under conditions of scientific uncertainty, see E Fisher, Risk Regulation and Administrative Constitutionalism (Hart Publishing 2007).
  11. HTAR, Art. 10.
  12. HTAR, Art. 10(5).
  13. HTAR, Art. 10(6).
  14. HTAR, Art. 3(7)(d).
  15. On the legal relevance of soft-law instruments, which national courts must take into consideration when deciding the disputes before them although they do not of themselves create rights, see Case C-322/88 Grimaldi v Fonds des maladies professionnelles [1989] ECR 4407, para 18.
  16. TIL-therapy completeness review (n 6), under Art. 9(3)(b)–(d) and 9(4) (Appendices A, B and D).
  17. On the appraisal of non-randomised evidence and indirect comparison see JAC Sterne and others, ‘ROBINS-I: A Tool for Assessing Risk of Bias in Non-randomised Studies of Interventions’ (2016) 355 BMJ i4919; and, on target trial emulation, MA Hernán and JM Robins, ‘Using Big Data to Emulate a Target Trial When a Randomized Trial Is Not Available’ (2016) 183 American Journal of Epidemiology 758. On population-adjusted (matching-adjusted and simulated-treatment) indirect comparisons specifically see DM Phillippo and others, NICE DSU Technical Support Document 18: Methods for Population-Adjusted Indirect Comparisons in Submissions to NICE (NICE Decision Support Unit 2016).
  18. catequentinib completeness review (n 7), under Art. 9(2) and 9(3)(a)–(b).
  19. Tovorafenib JCA (n 8): of the populations in the assessment scope, only the subgroup harbouring a BRAF V600E mutation could be assessed comparatively, by an indirect comparison against the dabrafenib-trametinib combination, the remaining populations being unassessable for want of comparative evidence.
  20. Implementing Regulation (n 5), Annex I (dossier and assessment-report templates); and Member State Coordination Group on Health Technology Assessment, Methodological Guideline for Quantitative Evidence Synthesis: Direct and Indirect Comparisons (2024), adopted under Art. 3(7), point (d), HTAR (n 1).
  21. Tovorafenib JCA (n 8): the report’s discussion of the certainty of the evidence and of the study limitations catalogues residual confounding, the limited comparability of the populations, small effective sample sizes and an uncertain proportional-hazards assumption, without assigning an overall graded level of certainty.
  22. GH Guyatt and others, ‘GRADE: An Emerging Consensus on Rating Quality of Evidence and Strength of Recommendations’ (2008) 336 BMJ 924-926; H Balshem and others, ‘GRADE Guidelines: 3. Rating the Quality of Evidence’ (2011) 64 Journal of Clinical Epidemiology 401-406.
  23. HTAR (n 3) Art. 9. The joint clinical assessment report “shall not contain any value judgement or conclusions on the overall clinical added value” of the technology and is “limited to a description of the scientific analysis of the relative effects” assessed on the chosen health outcomes.
  24. On proportionality as a general principle of EU law see T Tridimas, The General Principles of EU Law (2nd edn, Oxford University Press 2006) ch 3.
  25. Consolidated Version of the Treaty on the Functioning of the European Union [2016] OJ C202/47, Art. 263. On the distinction between reviewable and merely preparatory acts see Case 60/81 IBM v Commission [1981] ECR 2639, ECLI:EU:C:1981:264, paras 9–10; on measures producing binding legal effects, Joined Cases C-463/10 P and C-475/10 P Deutsche Post and Germany v Commission [2011] ECR I-9639, ECLI:EU:C:2011:656; K Lenaerts, I Maselis and K Gutman, EU Procedural Law (Oxford University Press 2014) ch 7.
  26. HTAR (n 3) Art. 13, under which Member States, when carrying out health technology assessment at national level, must give due consideration to the results of the joint clinical assessment; the provision is the principal mechanism by which the Regulation seeks to avoid duplication of the clinical assessment at national level.
  27. Consolidated Version of the Treaty on the Functioning of the European Union [2016] OJ C202/47, Art. 296. On the adequacy of reasons, including in technical and scientific assessments, see Case C-367/95 P Commission v Sytraval and Brink’s France [1998] ECR I-1719, ECLI:EU:C:1998:154.
  28. On the limited review of complex scientific and technical assessments see Case T-13/99 Pfizer Animal Health v Council [2002] ECR II-3305, ECLI:EU:T:2002:209, and Joined Cases T-74/00, T-76/00, T-83/00 to T-85/00, T-132/00, T-137/00 and T-141/00 Artegodan and Others v Commission [2002] ECR II-4945, ECLI:EU:T:2002:283.
  29. Charter of Fundamental Rights of the European Union [2012] OJ C326/391, Art. 20. On the comparability requirement inherent in the equal-treatment principle see Case C-127/07 Arcelor Atlantique et Lorraine and Others [2008] ECR I-9895, ECLI:EU:C:2008:728, paras 23 and 26.
  30. HTAR, Art. 31.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.
Copyright: This open access article is published under a Creative Commons CC BY 4.0 license, which permit the free download, distribution, and reuse, provided that the author and preprint are cited in any reuse.