How a device investigation differs from a drug trial
A device investigation plan differs from a drug protocol in how hard it is to write, in the guideline, templates, terminology and safety reporting its documents follow, and in the investigation itself: no phase I in healthy volunteers, a design that follows the device's risk class, outcomes that depend on the operator, analysis populations and adverse effects that involve the procedure as well as the device, harder blinding, and a writer with more influence in a smaller company. Some of the types of investigation that ISO 14155:2020 distinguishes are comparable to phases of drug development.
Difficulty compared with a drug protocol
One experienced writer holds that device protocols are much easier to write than pharmacological ones. In a pharmaceutical study everything can interact with everything, so the design has to be much more sophisticated, whereas a device usually acts mechanically, at least in its primary action, and has fewer possible interactions. Protocols are easy in an established area, and the challenging ones are those for a completely new area, which is rarely the case.
The investigation documents
The documents of a device investigation differ little from those of a drug trial: separate guidelines apply, but they offer similar guidance on good clinical practice, adapted to each product type. The important differences lie in the guideline, the templates, the terminology and the safety reporting.
| Aspect | Difference |
|---|---|
| Good clinical practice guideline or standard | ICH E6 for pharmaceuticals; ISO 14155 for medical devices and ISO 20916 for in vitro diagnostic medical devices |
| Templates for study or investigational documents | Mainly differences in structure and order of content |
| Terminology | Clinical study or trial, treatment, effect and so on for pharmaceuticals; clinical investigation, intervention, performance and so on for devices |
| Safety reporting | Differences in what to report and in reporting timelines |
Risk assessments are generally performed earlier for devices than for pharmaceuticals, and are in general more structured, with more procedures around risk, including the evaluation of residual risk.
The investigation itself
Clinical investigations are often less complex than drug trials and more adapted to the type of product. A medicinal product typically needs a standardised set of studies, from phase I in a small number of healthy volunteers, or in some cases severely ill patients, to phase IV post-marketing studies. A single investigation may be enough for a device if it provides clinical data that support all the claims stated for it.
The investigations differ from drug trials in the following ways.
- There are no phase I trials in healthy volunteers. Low-risk devices such as plasters or a tongue depressor need no clinical investigation. For high-risk devices such as surgical implants or a heart valve, interventional procedures in healthy volunteers would be unethical.
- The design of an investigation is tied to the device's risk class and intended purpose. Clinical data were previously often not needed for moderate-risk devices, but with the release of MDR 2017/745 they are increasingly required for many of them. The investigations those devices need are often less demanding than those for high-risk devices. Follow-up for a temporarily used device usually lasts only 30 days, to cover procedure-related events, and for an implantable device it usually spans several years.
- For a completely novel device with a novel implantation technique, outcomes and complications are hard to predict, because data from animal studies translate only partly into clinical practice and the technique may be refined along the way.
- Outcomes are operator-dependent when an interventional or surgical procedure is involved; the success of an artificial hip joint, for instance, clearly depends on the surgeon. This should be borne in mind when designing, analysing and interpreting an investigation. A novel surgical or interventional technique may involve a learning curve, so roll-in patients may be considered in the plan to cover at least part of it.
- Analysis populations are more complex. In an investigation of an implantable device, a patient in whom the procedure was started but abandoned, or who had the device explanted, raises the questions of which group they belong to and whether their follow-up would differ from that of other patients.
- Where a drug trial deems an event drug-related or not, an adverse effect in a patient with a device may be device-related or procedure-related. The distinction matters because the device itself can work well while its procedure is too complicated for some surgeons. When transfemoral transcatheter heart valves were developed, for example, the initial antegrade access route was too complicated, so the procedure was adapted to use retrograde access.
- Adverse event definitions themselves are basically the same as for drug studies. Device deficiencies, product issues that did not necessarily lead to an adverse event, also need to be recorded, because they might have led to adverse outcomes in less fortunate circumstances.
- Unlike in drug research, coding with the Medical Dictionary for Regulatory Activities (MedDRA) is rarely used, and disease-specific endpoint definitions, such as the Academic Research Consortium guidelines, may be more relevant.
- Devices have fewer possible interactions with the body than medicinal products, which can interact with body systems at the molecular level, so device investigations often need comparatively fewer patients.
- Blinding is more difficult because devices often differ in design, so often only single-blinded trials are possible, blinding the patients and possibly the core laboratory and the clinical events committee to the treatment. Placebo-controlled trials with sham procedures are very rare.
- Device companies are on average smaller than pharmaceutical companies, so the writer of a device investigation plan often has greater influence and more often contributes strategic insight than a writer of a clinical trial protocol for a large pharmaceutical company with standardised document development and highly specialised roles.
A writer with experience of pharmaceutical studies should be able to switch to device investigations easily, bearing these peculiarities in mind.
Types of investigation and the phases of drug development
Accounts differ on whether device studies have phases. One holds that they are not classified into phases I to IV as drug studies are, and that a variety of terms with similar meanings is used instead. Another holds that devices have phases just as drugs do, though the phases are not as clearly defined as for pharmaceuticals.
Before ISO 14155:2020 the types of clinical investigation were not clearly defined, and Annex I of the standard covers that gap. It distinguishes pre-market investigations, of devices that have not yet gained market approval (the CE mark in Europe), from post-market investigations, which follow market approval. For novel products in higher risk classes, first-in-human or feasibility studies may be needed to gain initial information on the device's safety and to determine whether the procedure is feasible, and these are comparable to phase II studies of medicinal products. Based on these studies, the device or the intervention may be modified as necessary, or new hypotheses inform the design and sample size of pivotal investigations, which are comparable to phase III trials. In the post-market phase an investigation may be interventional, where an intervention such as additional X-ray assessments occurs for the purpose of the investigation, or non-interventional, where patients are treated according to the standard of care at the facility. Company-sponsored and investigator-initiated investigations, and prospective investigations and retrospective analyses, are further approaches.
For phase II studies the nomenclature varies. Some call a study that enrols a very small number of patients, perhaps only three, to see whether the procedure is possible at all a feasibility or first-in-man study, but a first-in-man study can also be the one that leads to the CE mark. For a new drug-eluting stent, which can do little harm, the first study is usually the one that also leads to the CE mark. For a completely novel device such as a transcatheter mitral valve, where it is not known whether the device will need amending again, the first study enrols very few patients. The study that leads to the CE mark, called a pre-market study or by some a clinical study, and in the United States an investigational device exemption study, is equivalent to phase III. Device registries correspond to phase IV studies.