Clinical investigation reports for medical devices
A clinical investigation report describes the design, conduct, statistical analysis and results of a clinical investigation of a medical device. Writing one means knowing the standard that defines it, when it is written, who reviews, signs and receives it and when it becomes public, the inputs it is drafted from, how its structure maps onto a drug trial's clinical study report, what each part contains from the cover page to the annexes, and how long the work takes.
The standard and the regulation that define the report
ISO 14155:2011, Clinical investigation of medical devices for human subjects — Good clinical practice, is the device counterpart to ICH good clinical practice. It defines a clinical investigation report as a document describing the design, execution, statistical analysis and results of a clinical investigation, and its informative Annex D covers what the report contains. Clinical investigation report and clinical study report are two names for the same document, and writers use both.
The content of the report is described in ISO 14155:2020, and the minimum requirements are in Chapter III point 7 of Annex XV of the Medical Device Regulation (MDR), with further guidance in MDCG 2021-6.
ISO 14155 is not free, and in the year the MDR was released it cost under €100. One experienced device writer holds that a writer who needs it only for a clinical investigation report, and has its report requirements set out elsewhere, does not strictly need to buy it, but urges anyone continuing to work on medical devices to buy it. Annex D is informative, so a report need not follow it one to one, but staying as close to it as possible is good general guidance.
When the report is written and how often
A clinical investigation report is required at least once, after the investigation terminates, when it is closed and the statistical analysis has been performed, whether it ended regularly or prematurely.
A sponsor may want reports more often than that. A device study with five years of follow-up that is meant to lead to a CE mark does not wait for the follow-up to end, so its first report is usually written when the primary endpoint is reached, often after a shorter term such as 30 days, six months or one year. Companies quite frequently then produce a report every year, because the clinical study report also feeds into the clinical evaluation report. A manufacturer has to produce a clinical evaluation report regularly, to obtain a CE mark and then to maintain it, and it is a far more extensive document of which the clinical study report is an integral part.
Review, signature and distribution
The report is to be made available to the coordinating clinical investigator. In a multicentre study, the principal investigator at each site has overall responsibility at that site, and one coordinating clinical investigator oversees the principal investigators.
Under ISO 14155, the report should be circulated among the principal investigators for their comments. Ideally all principal investigators review it, and the final report should be reviewed by the principal investigators and the manufacturer. One experienced device writer finds that circulation too complicated in practice and, as long as nobody asks for the principal investigators' feedback, usually sends the report only to the coordinating investigator, who signs it. Under the MDR, the principal investigators' feedback on the report has to be sought.
The final report is signed by the sponsor and the coordinating investigator, or by the principal investigator in a single-centre investigation. The MDR adds the signatures of the principal investigators.
The signed report is made available to the ethics committee and/or the applicable regulatory authorities, depending on the country. It goes to the ethics committee, and to the regulatory authorities where the study required regulatory approval, within one year after study termination or within three months of a premature termination.
Submission and public availability
Under Article 77 of the MDR, the report and its summary for an investigation carried out for conformity purposes under Article 62, or on a CE-marked device under Article 74, should be submitted to every member state in which the investigation was conducted within one year of its end, although later submission may be justified for scientific reasons and specified in the clinical investigation plan. Where the investigation is terminated early or halted temporarily, the results should be published within three months. The report and its lay summary should become publicly available immediately after submission in cases of early termination or temporary halt, when the device is registered under Article 29 and before it is placed on the market, and at the latest one year after submission if the device has not been registered by then. MDCG 2021-6 further details the requirements and timelines for reporting clinical investigations.
The results of the investigation should also be published in a publicly accessible database, and used to update the risk analysis and the clinical evaluation.
Under the MDR, the clinical investigation report, together with its lay summary, is to be published online through the European device database. In the year the MDR was released, that platform was not yet live, and one experienced device writer expected it to take probably at least until 2020 to go live.
Once reports are public, one experienced device writer expects the need for very good and very solid reports to increase, because a document published online is different from one sent only to regulatory bodies.
Inputs to the report
The key inputs to the report are the clinical investigation plan, also called the clinical study protocol, the statistical output, meaning the results, with the statistical analysis plan, regulations and guidelines such as ISO 14155, and publications and presentations in the disease area. The clinical investigation plan supplies the introduction and background, the description of the investigational device and the description of the study design.
The introduction and background of the clinical investigation plan can be more or less copied into the report, but its quality should be checked first. That section is very frequently poorly written, because it is often produced by the company itself, and readers who find it copied into a report might associate the poor writing with the report's writer. How much that matters depends on how close the writer's relationship with the company is: one experienced writer, who has worked for some companies for six years, can afford to copy it, since those companies would take a weak section as copied from the protocol. Rewriting the section can be worthwhile where it can also serve the manuscript, and a well-written one may lead to the writer being asked to draft something for a manuscript next time.
The writer should ask for more than the tables of the statistical analysis. The statistical analysis plan is worth requesting, because the outcomes, or the way they were calculated, frequently do not follow it, and it is better to check. Line listings of adverse events go into an appendix anyway, and they also allow narratives in the results, such as the cases of patients who died or of device failure. Minimum and maximum values are worth requesting, not necessarily to report them, but as a check on whether the data were cleaned accurately, which very frequently they are not. A maximum body mass index of 270, for example, is impossible and is probably 27.0, and such an error can distort the analysis. It is good to have the mean, the median and the 95% confidence interval. One experienced writer has received results from 20-patient feasibility studies that gave only the mean, from a company that could not say whether the data were normally distributed. The statistician can also be asked for graphs such as Kaplan–Meier curves and box plots.
Checking the analysis should be the company's task, but device manufacturers are often start-ups or inexperienced, so it is good for the writer to keep an eye on it.
Once the statistical report is final and the report is being compiled, any questions about the statistical analysis are cleared with the statistician, and depending on the project the writer may also need to work with the data manager, the monitor and the clinical project manager.
The writer should be aware of disease-specific guidelines, such as the Academic Research Consortium guidelines or the European Society of Cardiology guidelines for a stent study. One experienced writer always recommends citing such guidelines or an expert consensus statement, because it makes the information much more valid: a statement made according to the guidelines of the European Society of Cardiology is stronger than one made without them.
The writer should also become familiar with the key publications and presentations in the disease area.
What the report covers and how it maps onto a clinical study report
The report summarises the clinical investigation plan, explains any deviations from it, and presents and discusses the results, with a critical appraisal of the results against the stated objectives. It must include data from all participating sites, so that no unfavourable data are excluded, and it must never reveal a subject's identity.
Annex D of ISO 14155:2020 is less structured and detailed than the clinical study report structure of ICH E3 and the CORE Reference, which allows a more flexible approach, but its main headings are fairly similar. Its sections are general, cover page, table of contents, summary, introduction, investigational device and methods, results, discussion and overall conclusions, abbreviated terms and definitions, ethics, investigators and administrative structure, signature page and annexes. One writer with clinical study report experience could use that experience to plan the report's structure and what to write in each section: its important subheadings, the level of detail, the cross-references to other sections or external documents, the tables or figures to include, the source documents to use and what to ask the subject matter experts.
Take a writer who has written clinical study reports but never a clinical investigation report, given no detailed template or example, for an investigation aimed primarily at comparing aripiprazole compliance with ABILIFY MYCITE against aripiprazole tablets alone in patients with schizophrenia. Section 6.1 of the report serves a similar purpose to section 9.4 of the clinical study report, describing aripiprazole and the device component of ABILIFY MYCITE. Section 6.2 could carry what sections 8 and 9 of the clinical study report carry, other than section 9.4, with the objectives, endpoints, assessments and overall design summarised using ICH E3 and the CORE Reference as a structural reference. ISO 14155:2020 designates only section 7 for the results, so ICH E3 and the CORE Reference could supply its level 2 and 3 subheadings for the disposition of subjects and the results of the primary, secondary efficacy, safety and other endpoints.
Cover page, contents and summary
Annex D, with the MDR's additions, asks the cover page for:
- the title of the clinical investigation
- the identification of the investigational devices, including names and models, and under the MDR the single identification number, the unique device identifier, for which no scheme yet existed in the year the MDR was released, and one was expected by 2020 at the latest
- where the title does not make them clear, a single sentence describing the design, comparison, period, usage method and subject population
- the name and contact details of the sponsor or the sponsor's representative
- the clinical investigation plan identification, usually the plan number, which many companies build from C for clinical investigation plan, the year and a sequential number, so that C1702 is the second study started in 2017
- the names of relevant parties, such as experts, biostatisticians and core laboratories
- a statement of whether the investigation was performed in accordance with ISO 14155 and any other applicable guidelines
- the date of the report
- the author or authors of the report
- the details and signature of the coordinating clinical investigator and of the principal investigator of each site
These requirements make for a crowded cover page, so one experienced writer puts on it only the title, the device name or a picture of the device, the sponsor details, the clinical investigation plan identification, a short statement that the trial was conducted according to ISO 14155 and applicable local regulations, and the date of the report, and moves the rest to the following pages.
The table of contents gives a page number for each section, including summary tables, figures and graphs, and a list of appendices.
The summary gives the title and purpose of the investigation, describes its population, design and methods, all drawn from the clinical investigation plan, then gives the key results, usually the endpoint results and perhaps some key baseline parameters, and the conclusion. Annex D lists the dates of initiation and completion last, and one experienced writer puts them at the start instead, as an introduction.
Introduction, device description and plan summary
Introduction. As in a manuscript, the introduction places the investigation in the context of the investigational device's development and relates the investigation's critical features to that development, such as its objectives and hypothesis, target population, treatment and follow-up duration. Under Annex D, it identifies or describes any guidelines followed in developing the investigation, such as exclusion criteria selected according to the European Society of Cardiology's recommendations, and any agreements or meetings between the sponsor and the regulatory authorities relevant to the investigation. One experienced writer has never seen the sponsor's discussions with regulators included in this section, and they need not be, because Annex D is informative.
Investigational device description. This section gives a very brief description of the device with pictures or drawings, its intended use, and any changes to the device during the investigation. Indication and intended use are different: for a stent, the indication is coronary artery stenosis in de novo lesions, and the intended use is to open the lumen of coronary lesions. Under the MDR the report is to be posted on a public website, so the section should not disclose too much detail or any confidential information about the device.
In medical device trials, especially early feasibility trials and those with complex or novel procedures, the device may well go through several iterations as the team learns it should be improved, and each iteration has to be described. It may even make sense to analyse outcomes before and after an iteration, as a post hoc analysis, and the differences are sometimes dramatic. In one transcatheter aortic valve trial, the valve-in-valve rate was fairly high at the start, and after a modification of the delivery system no further valve-in-valve procedure occurred, which such an analysis can show.
Summary of the clinical investigation plan. The report is a standalone document, and some people may read only it, so it summarises the plan, including any amendments with the rationale for each. It describes the objectives, the design, including study type and endpoints, and the ethical considerations. For the ethical considerations, a statement that the study was conducted according to ISO 14155 and applicable local regulations can serve. Monitoring is a new MDR requirement and is usually covered in the data quality assurance section. The summary also covers the subject population, the treatment and treatment allocation schedule, including how randomisation was conducted in a randomised trial, and concomitant medications and treatments, such as the dual antiplatelet regimen usually recommended for a set period in a stent trial. One experienced writer orders this material as design, then objectives, then exclusion criteria.
The summary specifies the duration of follow-up and the statistical analysis, including the hypothesis or pass–fail criteria, the sample size calculation and the analysis methods. Where the plan describes the statistics in a complicated way, the report can simplify them so that readers who are not statistically trained can understand them. A later publication demands still more compression, since it cannot give two pages to a sample size calculation.
Results
The results give the investigation's initiation date and, where applicable, its completion and suspension dates. A table of key dates can hold first patient in, last patient enrolled, the follow-up status while the study is ongoing, and the last patient visit once it is complete.
The results report the disposal of subjects and of investigational devices. A CONSORT flow diagram can serve for the subjects, and for a purely investigational device a sentence usually suffices, to the effect that all investigational devices were returned to the investigators and device accountability is complete.
The results then give subject demographics and baseline parameters, endpoint results and the performance analysis. Harms are reported as a summary of all adverse events, with a discussion of severity, treatment needed, resolution and the principal investigator's judgement of the causal relationship with the investigational device or procedure. One experienced writer gives a table of the numbers of adverse events, adverse device effects, serious adverse events, serious adverse device effects and device deficiencies, with the line listings as an appendix. Observed device deficiencies that could have led to a serious adverse device effect should also be described.
Subgroup analyses are added where appropriate, and a post hoc analysis is declared as post hoc, as in a publication.
Subject accountability includes a description of how missing data and deviations were dealt with in the analysis, which belongs in the clinical investigation plan and is usually forgotten. Device studies most frequently do not replace missing data, and a short statement that missing data are not replaced and the analysis is based on available data is then sufficient. Deviations can produce different analysis populations, such as the intention-to-treat and the per-protocol population.
Compliance with the clinical investigation plan is reported by summarising the protocol deviations, with their line listing as an appendix. A listing taken from the database holds only the per-patient deviations entered there, so an incorrectly stored investigational device, incomplete device accountability, or a device given to a patient who had not signed informed consent appears nowhere in it. One experienced writer therefore recommends obtaining written confirmation that no protocol deviation other than those in the line listing occurred.
Discussion and conclusions
The discussion and overall conclusion cover the safety and performance results and any other endpoints, an assessment of risk and benefit, and the clinical relevance and importance of the results in the light of other existing data, usually as a comparison with other published data. Under the MDR, the discussion also has to show that the device is state of the art, where before it was enough to show that it is safe and effective.
The report also lists any specific benefits and special precautions required, such as excluding patients with a pacemaker, any implications for the conduct of future clinical investigations, and any limitations of the investigation.
Closing sections and annexes
The remaining sections are terms and definitions, a list of all ethics committees consulted, the investigators and administrative structure, the names and addresses of the sponsor and its representatives, and the signature page. One experienced writer usually gives the list of ethics committees in an annex, with a single sentence in the body confirming that ethics approval was obtained at all centres before the study started. The investigators and administrative structure section briefly describes how the investigation was organised and lists the investigators with their affiliations and the names and addresses of any third parties, such as core laboratories, contract research organisations and consultants, and it can also be given as an annex. That writer departs from Annex D's order, ending on the results and discussion and placing the ethics statement and administrative structure earlier.
The annexes are the clinical investigation plan with its amendments, the instructions for use, the list of principal investigators and sites with copies of their CVs, the names and addresses of third parties, the list of monitors, the ethics committees, the tabulated data sets, meaning the line listings of plan violations, of all adverse events and of withdrawals and discontinuations, and, where an audit occurred during the trial, the audit certificate. In one experienced writer's experience, the sponsor usually collates them, since it holds the documents, and the writer provides the list of what they need to include.
Template, checklist, layout and consistency
A template is worth asking for, since most companies have one, which the writer can use and perhaps improve, and at least it gives something to work along. Most templates in this field are not available for free. In the year the MDR was released, one experienced writer knew of no clinical study report published online, though a search for clinical investigation report or clinical study report together with medical devices might return some hits, and expected templates to become plentiful once the European database went live. Failing a template, that writer recommends adhering strictly to the structure of ISO 14155 and turning its content requirements into headings. ISO 14155:2011 and the MDR can serve as a checklist of the content requirements.
The report should have a professional layout, because not every reader can judge its content but every reader sees how it looks: a report that looks good makes a good first impression, and one that looks like rubbish will probably lead readers to think its content is rubbish too. Publications are often written on the basis of the clinical investigation report, so a good report, especially for a new company, may lead to the writer being asked to write the clinical publications.
In one experienced writer's view, consistency and clarity matter most, in this and every other document. That writer has reviewed documents by people with five or more years of experience that contradicted themselves on pages 10, 20 and 30, and has seen absolute newcomers working from a template do a good job because they were thorough, had a clear line of thinking and were consistent. Being consistent, in that view, is already half the job done.
Time needed to write the report
How long a report takes is nearly impossible to say, because it varies with the data received, the complexity of the device and the study outcomes.
In the ideal case, with a company that supplies everything, a statistical output whose tables can be copied in and a clinical investigation plan whose introduction can be reused, a report can be written in a day. One experienced writer's fastest took under two days, for a company the writer had worked for for years, which knew what the writer demands and whose statistical output had been adapted, in a disease area where the writer is an expert. That writer's longest took about a week and a half, in an indication they knew less well, with both the data analysis and the clinical investigation plan in poor shape. That writer still advises scheduling more time than these, because the writer is fairly experienced, usually much faster than others, and has worked in the same disease areas for ten years. A completely new disease area needs time scheduled to become acquainted with it, and in that writer's view a newcomer should schedule one month. By that writer's own account, these estimates are not valid or accurate but give at most a very broad frame.
Writing a first report without device experience
One experienced writer frequently hears writers object that they lack experience, and holds that writing the report is completely doable, since writing a clinical study report is not rocket science. In that writer's view, a writer who is thorough, stays consistent, works through the content requirements as a tick list and makes sure everything is complete will be able to do it.