From a clinical question to a publishable one

Most papers that get rejected were not lost in the grammar, the statistics, or the figures. They were lost before the first sentence was written, at the question. A weak question cannot be rescued by good writing, clever analysis, or a high word count. A strong one carries a paper even when the prose is plain.

So we start where the work actually starts: turning something you are curious about in clinic into a question a journal will want to publish.

A topic is not a question

The single most common mistake I see in a trainee’s first draft is that it is about something rather than asking something.

“Platelet-rich plasma in tendinopathy” is a topic. It has no beginning and no end; you could write about it forever and prove nothing. Compare:

Does a single PRP injection reduce pain at six months, compared with saline, in adults with chronic Achilles tendinopathy?

That is a question. It names who, what, against what, and measured how. It can be answered yes, no, or the evidence is too weak to say, and any of those three is a paper. A topic can only be summarised. A question can be settled.

Before you do anything else, force your idea through this test: can it be answered? If the honest reply is “it depends what you mean,” you still have a topic.

Make it answerable: PICO

The fastest way to convert a topic into an answerable question is to break it into parts. For clinical work, the standard frame is PICO:

• P, Population: who, precisely? Not “patients with knee pain,” but “adults over 50 with radiographic knee osteoarthritis.”
• I, Intervention or exposure: what is done, or what differs?
• C, Comparison: against what? “Versus placebo,” “versus standard care,” “versus the other technique.” A question with no comparator is usually a topic in disguise.
• O, Outcome: measured how, and when? “Pain on a validated score at 6 months,” not “improvement.”

For studies that aren’t interventional, swap Intervention for Exposure (PECO) or drop the comparator for prevalence questions. The point is the same: a publishable question has named, measurable parts. When you can fill in every letter, you can answer it, and a reader can judge it.

Make it worth answering: FINER

Answerable is not enough. Plenty of answerable questions are not worth a journal’s pages. The classic test for whether a question deserves a study is FINER:

• Feasible, can you, with your data, time, skills, and mentorship, actually answer it? This is where trainees overreach. A perfect randomised trial you cannot run is worth less than a clean retrospective study you can.
• Interesting, would a colleague want to know the answer?
• Novel, does it add something not already known? (More on this below.)
• Ethical, can it be done without harm, with approval and consent?
• Relevant, would the answer change practice, policy, or the next study?

Feasibility deserves a hard, honest look before you commit months. Ask plainly: do I have access to the data or patients? Do I have the statistical help? Is there ethics approval, or time to get it? Who will supervise me when I am stuck? A good question that fails on feasibility is not yet your question, it is someone else’s, with more resources.

Find the gap, and prove it is real

Novelty is the criterion trainees most often assert and least often check. “No one has done this” is a claim, not a fact, and reviewers will test it in minutes.

Do the search before you fall in love with the idea. Run a focused look through PubMed and Google Scholar. Read the two or three most recent reviews on the topic, and read their final sections, where good reviews say explicitly what remains unknown and where the field should go next. That “future directions” paragraph is often a map of publishable gaps.

But a gap must be two things at once: real (genuinely unstudied or unresolved) and worth filling (the answer would matter). “No one has compared these two suture techniques in left-handed patients” may be real and still not worth filling. The strongest questions sit where a real gap meets a relevant consequence.

The one message

The chemist George Whitesides put it best: a paper exists to teach the reader one conclusion, not to display everything you did. So before you write, write your paper’s single sentence:

This study shows that ______.

If you cannot complete that sentence, you do not yet have a paper, you have a topic, or a pile of data. Everything in the finished manuscript exists to earn, support, and qualify that one sentence. The question you started with and that sentence should be two sides of the same coin: the question asks it, the sentence answers it.

Match the question to the study you can actually do

The shape of your question already decides the shape of your study, a question about effect needs a trial or a strong comparison; a question about how common needs a well-defined sample; a question about what the evidence says overall needs a systematic review. We will design backwards from there in Part 2. For now, simply do not ask a question your available data cannot answer. The most painful rejection is the one where the question is excellent and the design could never have addressed it.

The traps that sink trainees

Four patterns account for most early failures:

• The data-first paper. You have a dataset, so you hunt it for a question. This is backwards, and reviewers can smell it, the question feels retrofitted because it was. Let the question lead.
• The me-too. Repeating a known study with a different city and no new idea. Replication has real value, but only when you say why this replication matters.
• The salami slice. One study chopped into the thinnest publishable units. It inflates a CV and irritates editors; it rarely survives peer review intact.
• The unanswerable and the trivial. Too broad to settle, or so narrow and known that the answer changes nothing. Aim for the question that is just big enough to matter and just small enough to finish.

A worked example, from my own work

When my group wrote the review on the lymphatic system in musculoskeletal disease, the topic, “inflammation in joints”, was enormous and crowded. Thousands of papers study cartilage and the synovium. What almost no one had pulled together was what happens to the joint’s drainage, the lymphatic vessels that clear inflammatory debris, as disease progresses.

That was a real gap (few syntheses existed), and a relevant one (if drainage fails, restoring it becomes a target). The topic became a question: what does current evidence say about how the lymphatic system changes in musculoskeletal disease, and what does that imply for treatment? Answerable by a structured review, novel, feasible for us, and relevant. The one sentence wrote itself, this review shows that lymphatic dysfunction is a consistent, under-recognised feature of joint disease and a plausible therapeutic target, and the paper followed from it. It was published in Bone Research. The work that made it publishable happened at the question, not the keyboard.

Your turn, twenty minutes

Before you read Part 2, do this:

1. Write your topic in one line.
2. Break it into PICO (or PECO). If any letter is blank, your question is still a topic, fill it in.
3. Spend thirty minutes searching, and read the “future directions” of the newest review. Write the gap in one sentence: “What is not yet known is ______.”
4. Run FINER. Be ruthless about Feasible.
5. Write your one message: “This study shows that ______.”

If steps 3 and 5 don’t hold, iterate the question, not the prose. The hour you spend here will save you a month later.

---

Next in this series, Part 2: Choose your study type, and design backwards from its reporting guideline. Why the question you just wrote already tells you whether you need CONSORT, PRISMA, STROBE, or CARE, and how to build the study so the reporting writes itself.