Trial, RWE or scientific survey: choosing the right rigour path before you commit
Two symmetric mistakes cost sponsors the most: assuming a survey will cover everything, or building a full protocol for a question an agile instrument could have answered. The route is decided by the data you intend to capture, not by budget.
Insight · · 5 min
Two conversations happen almost every week, and they are mirror images of each other. In the first, a sponsor arrives convinced that a health scientific survey will cover the whole question: fast, light, no ethics committee, results in eight weeks. In the second, a sponsor arrives with a full interventional protocol for a question that was never causal to begin with, and will spend eighteen months and a seven-figure budget to learn something a well-built instrument could have told them in one quarter. Both mistakes are expensive. They are the same mistake in opposite directions: choosing the route before defining the data.
The route is not a budget decision. It is decided by the nature of the data you intend to capture and by what you intend to say afterwards. That is the whole of it. A survey stays a survey as long as it observes, with validated instruments, what people already do, already take, already perceive. The moment the design touches an allocation, an intervention, a biomarker drawn for study purposes, a clinical endpoint, or an outcome that will be used to support a claim on a product, it has left survey territory. At that point it needs ethics committee submission, health authority notification where applicable, informed consent to GCP standard, documented data provenance and an audit trail that survives a reviewer. Sponsors rarely intend to cross that line. They cross it in the design detail: one extra question about a prescribed therapy, one lab value, one before-and-after comparison on a marketed product, and the study is a study.
It helps to see the three routes side by side, not as a hierarchy of prestige but as a ladder of rigour.
A clinical trial is the route when the question is causal. You allocate, you control, you compare, and you can say the product caused the effect. It carries the heaviest governance: protocol, ethics approval, competent authority submission where required, GCP conduct, monitoring, safety reporting, statistical analysis plan, clinical study report. It is the only route that will support a strong efficacy claim, and it is the wrong route for anything that is not causal.
Real-world evidence sits in the middle. Retrospective, prospective or ambispective, registries, EHR-based research, observational programmes and patient support programmes. No allocation, no intervention: the clinical decision has already been made by the clinician. It still requires ethics review and, in most jurisdictions, a formal non-interventional study submission, because it handles identifiable health data on real patients. It answers questions a trial cannot: what happens outside inclusion criteria, over years, in the population that actually uses the product.
A health scientific survey is the lightest route, and the most misunderstood. Done properly it is not market research with a lab coat on. It uses psychometrically sound, validated instruments to measure habits, adherence, dietary behaviour, symptom burden, perceived benefit, unmet need, across consumers, patients and HCPs. When it is genuinely non-interventional, does not collect special-category data beyond what consent and design allow, and does not evaluate a product on identified patients, it can run without a protocol submission in many jurisdictions. Properly scoped, it produces publishable, defensible data.
The reverse case deserves as much attention as the first. A large share of what sponsors want to know is not causal at all. Why do people stop taking the product at week six? What does the pharmacist actually say at the counter? Which benefit do consumers recognise, and in what language? Where is the gap between the label and the lived experience? Is there a population worth designing a trial for at all? None of those questions need an interventional design, and a trial answers them badly. A well-built survey answers them faster, at a fraction of the cost, and, critically, sharpens the trial that comes afterwards: better endpoint selection, better inclusion criteria, better sample size assumptions, better claim language. Running the light route first is not the cheap option, it is the disciplined one.
Then there is geography, which is where most plans break. The threshold between an exempt survey and a study requiring approval is not uniform. Within the EU alone, national implementations differ on what counts as non-interventional, on whether HCP-directed research requires ethics review, on how special-category data under GDPR must be handled, on whether a local representative or a national registration is needed. The same questionnaire can be exempt in one member state and in scope in the next. A multi-country programme designed to the lightest national standard will stall at the first border where it does not hold, and one that is designed to the heaviest standard everywhere pays for governance it never needed. Neither is acceptable when the launch window is fixed.
This is precisely the work AIRA does before anything is built. It maps the evidence question against the data you intend to capture, the claim you intend to support and the jurisdictions you intend to run in, and returns the lightest defensible route per country, along with the submissions, timelines and dependencies each one carries. Not a recommendation to run the biggest study available, the opposite: the shortest path that still holds up. Where a survey suffices, it says so. Where a single data point pushes the design across the line into ethics territory, it flags which data point and what it would cost to keep it or drop it.
And nothing gets implemented on the strength of a route decision alone. Between the strategy and the field there is a design and configuration phase, run jointly with the sponsor's medical, marketing and regulatory teams in the same room. Medical owns the scientific question and the endpoint. Marketing owns the claim the evidence has to carry, and states it before the instrument exists, not after the data arrives. Regulatory owns the boundary conditions and the submission path. That phase is where the instrument is drafted, the population defined, the analysis pre-specified and the governance model agreed. Studies that skip it are the ones that discover in month four that the endpoint does not support the claim, or that the country they most needed required a submission nobody filed.
So the question to bring to your CRO is not which study is cheapest, nor which is most rigorous in the abstract. It is: what is the lightest route that still answers this question defensibly, in these countries, for the claim we intend to make? Get that right and the rigour stops being a cost. It becomes the reason the evidence is believed.
Written by the Evidilya scientific team. For interviews, references or a full publication list, use the contact page.
Talk to our team.
Bring us the evidence question you're wrestling with, we'll tell you the shortest defensible path.