From JCA Relative Effects to National Reimbursement:
Managing PICO Multiplicity, Indirect Comparisons and
Residual Evidence Gaps
The central technical challenge of the EU Health Technology Assessment Regulation is not simply producing a Joint Clinical Assessment (JCA) dossier on time. It is preserving the validity and decision relevance of one evidence package as it moves through two different systems:
an EU-level assessment of relative clinical effectiveness and safety; and
national appraisal processes that may add country-specific comparators, treatment pathways, economic models, budget constraints and reimbursement conditions.
The JCA creates a common scientific starting point. It does not create a single European value judgement, price or reimbursement decision.
That distinction has major implications for evidence strategy. A statistically credible treatment effect within one trial population and one comparator framework may not be directly decision-relevant for every Member State. Conversely, an analysis developed for one national authority may not answer the full set of consolidated PICOs examined at EU level.
Manufacturers therefore need more than a submission plan. They need an evidence architecture capable of translating relative effects across multiple populations, comparators, endpoints and national decision contexts without creating contradictory analyses or unsupported claims.
JCA harmonises the assessment process, not the decision context
Under Regulation (EU) 2021/2282, JCA is intended to provide a structured assessment of the relative effects of a health technology. Member States must give due consideration to the JCA report, but they retain responsibility for conclusions on added value, economic evaluation, pricing and reimbursement.
This creates an important analytical boundary.
The EU-level assessment asks whether the submitted evidence can support relative-effect conclusions across the assessment scope. National authorities may subsequently ask whether those effects are relevant to local practice, sufficiently certain for a particular reimbursement position and economically justified at the proposed price.
The post-JCA challenge is therefore not to repeat the same analysis country by country. It is to identify which parts of the EU evidence base are directly reusable, which require adaptation and which national questions remain genuinely unanswered.
Three forms of heterogeneity drive that translation problem:
Clinical heterogeneity: differences in baseline risk, disease severity, prior treatment, effect modifiers and outcome definitions.
Decision heterogeneity: differences in relevant comparators, positioning within the treatment pathway and evidentiary expectations.
Economic heterogeneity: differences in unit costs, resource use, discounting, time horizons, willingness-to-pay conventions and budget impact.
Treating these as one generic “localisation” exercise is a strategic mistake. Each has different implications for study design, statistical analysis and uncertainty.
PICO multiplicity is an evidence-network problem
Multiple PICOs are often described as a dossier-volume problem: more PICOs require more tables, analyses and narratives. The more consequential issue is whether the available evidence forms a connected and decision-relevant network for all requested comparisons.
For every PICO, the manufacturer should map:
the randomised evidence directly addressing the comparison;
the connected evidence network supporting indirect comparison;
single-arm, observational or registry evidence that may supplement the network;
effect modifiers that differ between studies or populations;
endpoint definitions and follow-up periods;
treatment switching, crossover and missing-data mechanisms;
the estimand addressed by each analysis; and
whether the analysis estimates an effect for the JCA population or requires transport to another target population.
This mapping should happen before statistical methods are selected. A network meta-analysis cannot solve a disconnected evidence network. A matching-adjusted indirect comparison cannot correct for unmeasured effect modifiers. A larger RWD source does not compensate for an unsuitable target-trial definition.
The correct question is not, “Which indirect comparison method should we run?
It is:
What causal contrast is required for this decision, in which target population, and what assumptions would make that contrast identifiable from the available data?
Start with the estimand, not the available dataset
Evidence translation becomes unstable when different functions estimate different treatment effects without recognising it.
Clinical development may focus on a treatment-policy estimand. A payer may be interested in the effect while patients remain on treatment. A national economic model may require long-term effects beyond trial follow-up. An indirect comparison may use an endpoint definition or assessment time that does not match the pivotal study’s primary analysis.
For every major analysis, teams should specify:
Population: Which patients does the estimate apply to?
Treatment condition: Initiation, adherence, switching and treatment duration.
Comparator: Trial control, current standard of care or a country-specific alternative.
Outcome: Exact endpoint definition and measurement schedule.
Intercurrent events: Treatment discontinuation, rescue therapy, switching and death.
Summary measure: Hazard ratio, risk ratio, restricted mean survival time, mean difference or another measure.
Time horizon: Trial follow-up, a fixed landmark or extrapolated lifetime effect.
An “overall survival benefit” is not one interchangeable quantity. A hazard ratio obtained under proportional-hazards assumptions, a restricted mean survival-time difference and a modelled lifetime survival gain can lead to different interpretations and different economic results.
The evidence plan should make these differences explicit rather than hiding them behind a single value narrative.
Choosing among indirect comparison methods
When the national comparator was not included in the pivotal trial, the analytical route depends on network structure and data access.
Network meta-analysis
Network meta-analysis is generally appropriate when a connected network of randomised studies exists and the transitivity assumption is credible. The assessment should not stop at a comparison of inclusion criteria. It should examine the distribution of plausible effect modifiers across studies, changes in background care, endpoint ascertainment and follow-up.
Where networks contain few studies, heterogeneity estimates may be unstable. Fixed-effect models can appear precise while underrepresenting genuine between-study variation; random-effects models can be weakly estimated. Scenario analyses should therefore test alternative heterogeneity assumptions and network structures.
Population-adjusted indirect comparisons
Matching-adjusted indirect comparison (MAIC) and simulated treatment comparison (STC) may be considered where individual patient data are available for at least one treatment but only aggregate data are available for the comparator.
Their credibility depends on identifying and adjusting for all relevant effect modifiers and prognostic variables. In an unanchored comparison, the assumptions are substantially stronger because prognostic imbalance must also be addressed. Effective sample-size loss after weighting is not merely a reporting statistic; it indicates limited population overlap and should affect the confidence placed in the result.
External control arms
External controls constructed from registries, electronic health records or other RWD may be useful where randomisation is infeasible or the available comparator evidence is inadequate. However, they require a clearly emulated target trial:
aligned eligibility criteria;
a common time zero;
consistent treatment assignment definitions;
harmonised outcome measurement;
control of measured confounding;
assessment of immortal-time and selection bias; and
sensitivity analyses for residual confounding and missingness.
Without these design elements, sophisticated adjustment can produce a precise estimate of the wrong contrast.
RWE should answer a defined residual gap
RWE is most valuable after the residual evidence gap has been characterised precisely. It should not be added to a dossier simply to demonstrate that “real-world data were considered.”
Potential post-JCA uses include:
estimating baseline event rates in national clinical practice;
quantifying treatment patterns, sequencing and comparator displacement;
assessing whether trial participants represent the national target population;
estimating resource utilisation and treatment duration;
validating surrogate-to-final outcome relationships;
characterising rare safety outcomes;
informing conditional reimbursement or managed-entry agreements; and
updating model inputs as clinical practice evolves.
For causal relative-effect estimation, a fit-for-purpose assessment should evaluate:
Relevance: Does the data source contain the target population, exposure, comparator, outcome and follow-up required by the research question?
Reliability: Are data provenance, completeness, linkage, coding and quality controls sufficiently documented?
Design validity: Does the study emulate the protocol of the hypothetical target trial?
Analytical validity: Are confounding, selection, measurement error, missing data and censoring appropriately addressed?
Transportability: Can the estimated effect be applied to the population relevant to the national decision?
The most important discipline is to separate three tasks that are frequently conflated:
estimating a treatment effect;
transporting that effect to a different population; and
extrapolating that effect beyond the observed follow-up.
Each requires different assumptions. Combining them in one model does not eliminate those assumptions; it makes them harder to see.
Transportability must be demonstrated, not asserted
A JCA-relevant treatment effect may not be directly transportable to a national population if effect modifiers differ materially between the trial and routine care.
Teams should compare the trial, JCA and national populations across:
age and disease duration;
severity and risk distribution;
prior and subsequent therapies;
comorbidities and concomitant medication;
biomarker prevalence;
adherence and persistence;
healthcare setting; and
availability of diagnostic or monitoring infrastructure.
Standardised differences can describe imbalance, but they do not establish transportability. The analysis must identify which variables are plausible effect modifiers rather than adjusting indiscriminately for every observed difference.
Where individual-level target-population data are available, reweighting or outcome-modelling approaches may support transport. Doubly robust methods can reduce dependence on one model specification, but they still depend on conditional exchangeability, positivity and consistent measurement.
Where overlap is weak, the honest conclusion may be that the effect is not reliably transportable. That is a decision-relevant result, not an analytical failure.
Relative-effect uncertainty must remain visible in the economic model
National reimbursement often introduces a second transformation: clinical relative effects become quality-adjusted survival, costs and budget impact.
This step can amplify uncertainty. A modest difference in the assumed duration of treatment effect may materially change the incremental cost-effectiveness ratio. A hazard ratio treated as constant over a lifetime horizon can create implausible survival projections. A local comparator absent from the trial may require both indirect comparison and long-term extrapolation.
Clinical and economic teams should jointly test:
proportional-hazards versus non-proportional-hazards models;
treatment-effect waning and stopping rules;
alternative parametric and flexible survival functions;
structural uncertainty in health-state definitions;
correlation between treatment effects and model parameters;
alternative comparator market shares and displacement assumptions; and
scenarios using JCA-wide versus country-specific populations.
The objective is not to generate one model that produces an acceptable result in every country. It is to preserve traceability from each national economic output back to the clinical estimand and evidence source on which it depends.
A practical evidence-translation matrix
Before the JCA report is finalised, manufacturers can build a country-level translation matrix.
| Evidence component | EU-level question | National translation question | Likely analytical response |
|---|---|---|---|
| Population | Which populationsare | Does the reimbursement population | Subgroup analysis, |
| Comparator | Which comparators | Is the locally relevant comparator directly represented? | NMA, MAIC/STC, external control or explicit evidence-gap statement |
| Outcomes | What relative clinical | Are country-specific utilities, resource use or final outcomes required? | Outcome mapping, RWE inputs and economic modelling |
| Follow-up | What duration is observed | Is extrapolation required for reimbursement modelling? | Survival extrapolation, waning scenarios and validation |
| Uncertainty | What limitations are identified in the JCA? | Which uncertainties are decision-critical nationally? | Scenario analysis, additional evidence generation or managed-entry agreement |
| Implementation | Whatclinical positionissupported by the evidence? | How will the technology enter thelocal pathway and | Pathway mapping, uptake scenarios and budget-impact analysis |
This matrix should be version-controlled and linked to an evidence-generation roadmap. Every national request should be classified as:
already answered by the JCA evidence package;
answerable through adaptation of an existing analysis;
dependent on a new analysis using existing data; or
dependent on new data collection.
That classification prevents duplication while making genuine residual gaps visible early.
Governance matters as much as methodology
PICO multiplicity becomes unmanageable when global clinical, EU HTA, HEOR, statistics, epidemiology and national affiliates maintain separate evidence inventories.
A defensible operating model requires:
one controlled repository of studies, datasets and analytical outputs;
a common estimand register;
a comparator and evidence-network map;
prespecified rules for selecting direct, indirect and RWE analyses;
documented assumptions and sensitivity analyses;
clear ownership of EU-level versus national adaptation; and
change control when the label, JCA scope or national pathway evolves.
The aim is not complete uniformity. Country adaptation is unavoidable. The aim is analytical coherence: different outputs should be traceable to explicit differences in decision context, not to uncoordinated modelling choices.
The strategic objective: one evidence architecture, multiple valid decisions
JCA changes the sequencing of evidence generation. Manufacturers can no longer wait for an EU assessment to finish before asking what Germany, France, Italy, Spain, the Netherlands or other markets will require. By then, the most important evidence limitations may be fixed in the trial programme and the remaining timelines may only allow analytical adaptation, not new evidence generation.
The strongest strategy is therefore neither one universal analysis nor dozens of isolated country dossiers. It is a modular evidence architecture built around:
clearly defined estimands;
an explicit PICO-to-data map;
connected and critically assessed evidence networks;
fit-for-purpose indirect comparison methods;
RWE studies designed for specified residual gaps;
transparent transportability and extrapolation assumptions; and
early integration of clinical and economic uncertainty.
The organisations that manage this translation layer effectively will be better positioned to reuse JCA outputs, respond to national requirements and defend the consistency of their value proposition across Europe.
The future of European Market Access will not be determined by how many analyses a company can produce. It will depend on whether each analysis estimates the right effect, for the right population, against the right comparator, and whether the uncertainty remains visible when that evidence reaches the reimbursement decision.
Continue the technical discussion in Amsterdam
The 2.0 European RWE & Market Access Summit 2026 – JCA & PICO Strategy Edition will bring together senior leaders from Market Access, HEOR, RWE, Pricing and Reimbursement, EU HTA and evidence generation to examine how organisations can translate European assessment requirements into defensible national access strategies.
21–22 October 2026
Inntel Hotels Amsterdam Landmark, Amsterdam
View the European RWE & Market Access Summit
Join senior professionals from pharmaceutical and biotechnology companies, HTA bodies, payer organisations, consultancies, data partners and evidence-solution providers to examine how Europe’s new assessment environment is changing evidence generation and access strategy.