Writing for a regulator or assessor you will never meet
An agency's reviewer or a notified body's assessor reads a regulatory document in order to decide on it, and many writers never meet that reader. A writer can learn about that reader from the documents the agency makes available. The authority, not the company, is the document's end client, and it wants the conclusion first, the evidence where its reviewer looks, a submission concise but complete and open about its weaknesses, and cross-references only where they add information. A device submission tells the device's story, with the patient at its centre.
Learning about an agency from its documents
Many writers learn as their first rule to know the audience, but what that means is unclear when the audience is the United States Food and Drug Administration (FDA): many medical writers who prepare documents for the FDA will never meet their reviewers, and it can be a struggle to learn who the audience there is.
The FDA is driven and bound by the Code of Federal Regulations and its own guidances. The regulations are the rules governing the agency, while the guidances represent its current thinking on particular topics and so are not legally binding. Many writers are likely to know these, but they may not know the FDA's Manual on Policy and Procedures (MAPP), which sets the standards for the regulatory documents the FDA itself prepares. The MAPP details how the FDA structures its summary reports and its decisions on the drugs it reviews. Comparing sponsor-submitted documents with the MAPP can reveal sections of the former cut and pasted into the latter, along with the reviewer's comments.
In the view of a writer who has attended many FDA review meetings, the documents the FDA makes available online are the best way to begin to know it. Starting with the codes and guidances most relevant to the writer's own work is a well-known beginning, but it is only the tip of the regulatory iceberg. As of 2013, the Drugs@FDA section of the agency's website held general information on competitor drugs and many of the documents in their approval dossiers, which show how the arguments for approval were constructed. Besides the expected documents, it held the minutes of any pre-new drug application (pre-NDA) meeting, comments from discussions and administrative review, and advisory committee documents. These documents can give unique insight into the thinking of the FDA and the sponsor on efficacy and safety testing. The pre-NDA minutes can be especially informative, because the meeting is designed to help resolve the outstanding questions and issues that have arisen during the clinical evaluation of a new drug.
The authority as the reader who decides
Regulatory documents are submitted to health authorities such as the FDA and the European Medicines Agency to get a drug, biologic or device approved for market, and they are written specifically for regulators, not for lay audiences. In one consultant's view, the documents are ultimately aimed at the regulatory authorities, and the authorities, not the pharmaceutical company, are the end clients, whether the writer works for the company directly or as an external consultant. The consultant believes that more dialogue between medical writers and the authorities would give everyone a much better understanding of what the authorities need and of how to provide the information in the format they want and need to assess medicines.
Scientific writing is inductive: it moves systematically through introduction, methods and results to the conclusion at the end. Regulatory writing is deductive: it starts with the conclusions and places the most important information up front. Scientific writing is descriptive and very detailed, with a data-focused presentation of the results, while regulatory writing is message-driven, focusing on the data that support the sponsor's specific position, and issue-focused, anticipating the reader's specific questions. A regulatory document is non-promotional and message-driven, and it needs to support the health authority's decision on whether to let a trial proceed or to approve the drug. It is not an academic discussion or a display of technical excellence, but is about getting the drug approved.
The audiences differ too: the reader of scientific writing is an expert, excellent in a narrow field, while the reader of a regulatory document has broad regulatory knowledge and is being helped to make an informed decision. To help the agency decide, the writing has to be visually appealing, really concise, lean and strategic, and very clear about the interpretation of the data. So in nonclinical writing the writer has to prioritise the most relevant information, and does so through ease of navigation: bookmarking, hyperlinking and a style written for reading on screen.
Regulators are experts at reviewing data: they see a lot of it and know what they want to see. They will still come back with questions, which can lead to delays, so everything has to be clear. They are not experts on the sponsor's compound, so the writer has to educate them on what the molecule is and what it does.
Where the reviewer looks
Regulatory writing aims to convey the message from clinical trials accurately and concisely, in a way that complies with regulations, to regulators who have a deep scientific background but are incredibly busy. The documents are highly templated, divided into sections with specific information expected in each, and regulators hardly ever read one from beginning to end, because they do not have the time. The aim is that a regulator with expertise in a field can go straight to the section where the information should be. In one regulatory writer's account, a regulator who does not find it there puts the document to the bottom of the pile and turns to documents written the correct way, so following the templates and putting the correct information into each section is very important.
FDA approval times had shortened, and in 2019 its reviewers were under pressure to assess more applications in less time. One regulatory writer holds that regulatory writers have a unique opportunity to influence the design and message of regulatory documents, which may ultimately help get important medicines to patients faster by relieving health-authority reviewers of unnecessary burdens.
Health authorities have very limited time, so a regulatory document has to be clean and lean, not repetitive, and focused. Its content has to be well structured, down to the sentences. A description of a drug's safety can repeat one sentence structure and be boring, because what matters is that the reader understands the meaning quickly, on the first read. Navigation through the sections and between documents should also be easy.
Reading on screen
Telling the story to a regulatory agency, as many writers learned to do, implies a reviewer who reads a document in order. As of 2019, busy reviewers actually read on screens, and documents must be designed for that, with keywords, scannable subheadings, navigational links, strategically placed white space, paragraph chunking and less dense text. Designing for the way reviewers read on screen, not on paper, should guide writing and development decisions, such as placing hyperlinks strategically and using the left side of the screen for focused keywords and subheadings. Eye-tracking of people reading on websites found an F-shaped pattern, in which most readers read the first paragraph, skip the middle and go on to the second paragraph. One nonclinical regulatory writer holds that regulators read documents this way, and that the first thing a regulator does is read the very first paragraph, so the most important information and the data belong there.
A paragraph should open with its key message, followed by the details that support it. This deductive model keeps the reviewer focused on the message first, without getting bogged down in details, which can cause reviewers to skip key information. To fit the way reviewers read on screen, a regulatory writer can:
- split up long paragraphs, with one idea per paragraph
- put important information and key messages first, followed by supporting information
- link to exhaustive analyses instead of reproducing them in the document
- use short, active-voice sentences
- use tables and figures to summarise complicated data
- place links strategically so that reviewers can focus on key messages
Links to tables or other sections should follow the key message rather than precede it. A sentence such as "Tables X, Y, and Z may be found in Section 15.2.3" belongs at the end of a section of a clinical study report, not at its very beginning, so that the reviewer can focus on the key messages first. Too much information can be damaging: an expert's "Go ahead and keep it in—it couldn't hurt" more often than not signals ancillary details that can detract from the key information.
Drafting from the key messages
Regulatory writing groups commonly complain that documents are written without a plan, that there is never enough time, that there are too many reviews and reviewers, and that people rewrite what has already been wordsmithed. One regulatory writer reports good experience with teams crafting the key messages outside the final document, on PowerPoint slides. Seeing content outside a document format lets a team think differently about it. Teams usually have an idea of the desired study outcomes even before database lock or the end of a stability study, so the key statements can be drafted from base-case scenarios and the data filled in later. The regulatory writer can guide the team through the approach in six steps:
- Align on the key messages in a "shell" PowerPoint slide, with examples and instructions for team members to fill in, and with bullet points for supporting messages to be used as new topic sentences, as in the key message "WonderDrug 10 mg, given orally twice daily in subjects with Disease who are on standard of care, significantly reduced Symptoms, with no increased safety risk, compared with placebo/active."
- Transfer the bullet points as they are to the document shell, determine their placement, flesh out the paragraphs and check text density, aiming for at least three to four paragraphs a page.
- Determine the logical architecture: which details can go into tables, and whether all the results need to be up front or can be placed later in the document.
- Review the document for searchable words a regulatory reviewer would look for, checking whether key statements begin with the keyword, in the active voice, and whether a sentence can be rearranged to bring the keyword to the left of the screen, as in "Cardiac toxicity (including pericardial effusion, hypotension, and cardiac ischemia/infarction) was balanced across treatment arms."
- Make sure the navigation works, with table and figure titles that reflect what is in them and are succinct rather than overly detailed, and subheadings that give navigation prompts on the left side.
- Paste the key messages, which should be the first sentence of each paragraph, into a blank document and compare them with the original slides for anything missing, lost information or creeping additions.
A team that developed an FDA briefing package this way reported that reviewer feedback was extremely positive.
Concision and disclosure
A writer should go beyond understanding the audience and get into the shoes of the notified body or health authority, whether the submission is for a standalone device or a combination product, asking whether they would want to read 500 pages of dense and convoluted text. Many people think that adding detail makes a submission or document safer, but in one practitioner's opinion it does the opposite: it shows poor mastery, it is prone to more errors and inconsistencies, and it increases the risk of questions. So a submission should be concise but complete, because concise is not only clearer but also safer. The writer should also be very firm in representing the negative aspects.
In the experience of a writer who has attended many FDA review meetings, the odds at a pre-NDA meeting of a friendly review, rather than a committee lying in wait to pounce, can be improved by a well-written dossier and also by a full, clear disclosure of the drug's strengths and weaknesses. Trying to minimise a drug's weak points is not a good idea, because the FDA is expert at ferreting out the deficiencies in applications. Burying or hiding information will only hurt the writer and the client, while a trusted reputation helps them obtain a fair review the next time.
Cross-referencing
Regulatory documents are often over-cross-referenced, in one writer's impression, because writers err on the side of caution. They fall over themselves to make sure regulatory reviewers find what they want, not wanting to be accused of failing to follow the templates or, worse, of hiding unfavourable results. The attitude is commendable, but it can leave a document littered with cross-references that give the reader no further information. As a rule of thumb, a cross-reference to a source that adds no further information should be omitted. In the section presenting the main efficacy results of a Summary of Clinical Efficacy, for example, a cross-reference to the subgroup analyses for a particular variable may be appropriate, while one in the opposite direction would probably not be useful, because the main efficacy results add no extra value to the subgroup analyses. A reference to a reference that then leads to an appendix is guaranteed to annoy.
The device's story
When asked what makes for good regulatory writing, senior regulatory experts often use the language of storytelling, and 12 such experts, when interviewed, answered with some version of "It all depends on how you tell your device's story." Longtime insiders who have worked for the FDA or notified bodies tend to stress the writing. By story they meant an organising principle, not narrative in the sense of the medical humanities. Their literary vocabulary of setting, plot and character, amplified with particulars extrapolated from their generalities, covers the practices good device storytelling needs.
Setting is context, and when asked how to tell a device's story, the experts usually said first to provide enough context, but not too much. Too many details overwhelm, so the writer should strive for the right ones. Reviewers of regulatory submissions often have impressive technical backgrounds but will likely know little about the device, how it was developed or how it fits into a specific medical field. Prejudices from previous experience might incline them to unhelpful assumptions about the device, and pre-empting such misunderstandings is the writer's job. Putting the device in the appropriate setting makes the submission believable.
Two practices can help establish the setting. The first is to make use of executive summaries and conclusions rather than treat them as unimportant padding, because boilerplate in place of passages crafted to convey the context throws away a vital chance to communicate with the reviewer. The second is attention to phrasing, as in a paragraph from a pre-submission letter to the FDA, altered and with data removed, whose aim was to show that a procedure off-label in the US was safe and had clinical data, although from elsewhere. Opening the paragraph on "Other US surgeons" signals that the discussion is still about the US, and placing a quoted surgeon in a US city reminds the FDA reader that the opinion comes from US-based surgeons. Closing on "Due to its recent development, long-term data is not yet available" fixes the data in time as well as place, and together these small additions create an interpretive framework for the reviewer.
Plot is causal connection, in E. M. Forster's sense: "The king died and then the queen died" is a story, but "The king died and then the queen died of grief" is a plot. Almost all the experts interviewed stressed connection between claims and data, between arguments and between sections, and this applies even to long documents such as the clinical evaluation reports required as part of marketing submissions in the European Union. Where explicit connections are missing and the reviewer struggles to follow the case, the reviewer may give up in frustration and write the applicant off as unprepared, or start to wonder what is not being said. A lack of clear causal connections can erode the writer's authority and invite counterarguments that might never have arisen with tight transitions between points.
Rationales should not be skipped. Robust rationales are critical to communicating the logic behind design changes to the regulatory agency, and the principle of explaining why something changed rather than glossing over how it changed can be applied to all the claims in the documentation. Other strategies can help keep the causal connections clear: signal terms such as first and second or after and before, for a sense of order; transition words such as another, yet, while and furthermore; reminders of the document's overall arc and how each point relates to it; topic sentences used consciously even inside a template; and sentences that proceed from a familiar subject to a new predicate, so that each links to the next. A paragraph from a clinical literature review on a device used in high-risk percutaneous coronary interventions shows the last of these: it runs from the procedure's early days, when success was below 60% and a cardiothoracic surgeon had to be in the operating room, to the narrower present definition of high risk, and each sentence's subject picks up the previous one's predicate. Regulatory documents should be written in Plain English, sometimes called Global English, whether they are for a notified body or the FDA. Plain English is simplified and strives to be nonidiomatic, and attention to the subject–predicate structure of sentences can help produce its clarity.
Character is focus, and the true protagonist of every regulatory document is the patient. The patient should be kept in mind whatever the document, even one not intended for the patient, because the patient will receive the device. Each piece of data represents an actual person. Reviewers at regulatory agencies are there because they want to help people, both by keeping them safe and by enabling new developments, so tone and terminology should not lose that human focus. Other characters matter too: the device may be the focus, but related alternative treatments are key to how readers perceive its value. Establishing the setting, keeping the causal connections clear and maintaining the focus on key characters make a good start.