Questions to Ask in a Journal Club
For residents, nurses, pharmacy students and graduate students who attend or present at a journal club and want something sharper to say than 'interesting paper'. The questions follow the order a session usually takes: what the study set out to ask, how it was designed and where bias could get in, a short group for papers that are not trials, what the numbers show, whether any of it applies to your own patients or work, and what to put to the presenter and the room. The guide after the list covers how to ask questions in journal club without derailing the session, whether you are in the audience or at the front.
Want questions from the whole vault instead? Try the random question generator.
The questions
Each question, and why to ask it
The study question
What question was this study trying to answer, in one sentence?
Why ask it
Settle this before anything else, because every later criticism depends on it. For a clinical paper, try filling in four blanks: who was studied, what was done to them, what it was compared with and what was measured. It is also the safest first question for a newcomer, since it needs no statistics.
What was already known, and what gap did the authors say they were filling?
Why ask it
The introduction is where authors make the case that their study was needed, so it deserves the same doubt as the results. Someone in the room may know the earlier work well enough to say whether the gap was real. When nobody does, that is a fair thing to look up before the next session.
Is this the right kind of study for this kind of question?
Why ask it
Whether a treatment works, what causes a disease, how accurate a test is and what an illness is like to live with each call for a different design. Ask what the ideal study would have looked like, then what stopped the authors from running it. Cost, ethics and time are honest answers.
What was the primary outcome, and was it chosen before the study started?
Why ask it
Look for a registration number or a published protocol and set it beside the paper. If the main outcome changed between the two, raise it as a question about what happened, since there can be a stated reason. A paper with no registration at all leaves you unable to check, which is worth saying aloud.
Does the main outcome measure the thing we care about, or a stand-in for it?
Why ask it
Lab values and scan findings move faster and are cheaper to count than deaths, admissions or how people feel, so trials often measure them. The follow-up is what would have to be true for a shift in the marker to mean a better life for the person, and whoever in the room treats the condition will usually have a view.
What was the comparison group given, and is that what we would actually do?
Why ask it
A new treatment can look good against a placebo, a low dose or an approach nobody uses any more. Have the control arm described in as much detail as the treatment arm: the drug, the dose, how often people were seen. Where it differs from usual care in your setting, the size of the benefit may not carry over.
Design and bias
Who was included, who was excluded, and how many were screened to find them?
Why ask it
The flow diagram answers this faster than the methods text, so ask for it on screen. A trial that screened thousands to enroll a few hundred may have studied a narrow group, and the exclusion list shows which of your own patients were never in it.
How were people assigned to the groups, and could anyone have known what was coming next?
Why ask it
These are two separate things: how the sequence was generated, and whether it was hidden from the staff enrolling people. Ask the presenter to point to the sentence that covers the second. If there is none, the fair summary is 'not reported', which is different from 'not done'.
Were the groups alike at the start?
Why ask it
This is the baseline table, usually Table 1. Skip any p values in it and look for differences large enough to matter to the outcome, such as age, severity or a prior event. Then ask the room which single imbalance would worry them most and which way it would push the result.
Who knew which treatment each person was getting?
Why ask it
Go through the list: participants, the staff treating them, the people recording outcomes, the analysts. Blinding matters most when the outcome involves a judgment, such as a pain score or a decision to admit, and least when it is death. A sharp follow-up is whether side effects would have given the game away.
Apart from the treatment being tested, were the two groups cared for in the same way?
Why ask it
Extra clinic visits, more phone calls or a second drug given more often in one arm can produce a difference that has nothing to do with the treatment under study. The methods say what was meant to happen, and the results or the supplement say what did. It matters most in trials where staff knew who was getting what.
How was the outcome measured, and was it done the same way for everyone?
Why ask it
Outcomes pulled from records, recalled by patients or judged by a clinician each bend in a different direction. Find out who did the measuring and whether they knew which group the person was in. If the outcome is a score or a questionnaire, check whether it had been tested before or was built for this study.
How many people dropped out or were lost to follow-up, and was it even across the groups?
Why ask it
Compare the number randomized with the number in the final analysis for each arm. Uneven loss is the larger worry, because people can leave for reasons tied to the treatment. A useful test for the room: would the result survive if everyone lost from the treatment arm had done badly? Some papers run that check themselves as a sensitivity analysis.
Were people analyzed in the group they were assigned to, whatever they ended up receiving?
Why ask it
That is an intention-to-treat analysis, and it protects what randomization built. A per-protocol analysis answers a different question: what happens to those who took the treatment as planned. The paper should say which one is its main result, and a gap between the two deserves a few minutes of the room's time.
Was the follow-up long enough for the outcome, and the harms, to show up?
Why ask it
Benefits and harms can run on different clocks, so a trial can be long enough for one and too short for the other. Before looking at the paper's figure, ask the clinicians present how long they would want. Then compare.
How did they decide how many participants they needed?
Why ask it
The sample size paragraph states the difference the authors were counting on, and the room can judge whether that was realistic or hopeful. When a small study finds nothing, the fair conclusion is that it could not tell, which is a long way from showing the two options are the same.
Did the study run as planned, or was it stopped early or changed partway through?
Why ask it
A trial halted at the moment a difference looks large may have caught a chance high point, so what matters is whether the stopping rule was written down in advance and who made the call. Changes to the protocol tend to be listed in the supplement, which someone should open before the session ends.
Who funded the study, and what part did the funder play in running or writing it?
Why ask it
The answer is usually printed near the end, under funding and disclosures, and takes seconds to find. Funding alone does not sink a paper. What matters more is who held the data, who ran the analysis and who had the final say on publishing.
Other study types
In an observational study, what else differed between the groups besides the exposure?
Why ask it
Name one specific candidate, such as smoking, income or how sick people were to begin with, and ask whether it was measured and adjusted for. 'They adjusted for confounders' is not an answer until someone reads out which ones. Whatever was never measured cannot be adjusted away, and saying so is fair.
If this is a study of a test, what was it checked against, and did everyone get that check?
Why ask it
A diagnostic study is only as good as its reference standard. When only the people with a positive result went on to the definitive test, the new test can look better at catching disease than it is. Ask too whether the person reading the reference test already knew the first result.
For a systematic review or meta-analysis, how did they search, and how alike were the studies they pooled?
Why ask it
Three things to hear from the presenter: which databases were searched, whether unpublished work was sought and who judged the quality of each study. Then look at the forest plot. If the individual results point in different directions, a single pooled number may be hiding a disagreement worth discussing.
With a qualitative paper, who was interviewed, and how did the researchers get from the transcripts to their themes?
Why ask it
The details to listen for are who was invited and who declined, who did the interviewing, and whether more than one person coded the transcripts. Look for quotations that complicate a theme as well as ones that support it. Sample size is the wrong yardstick here: the fair test is whether a reader can follow the path from what people said to what the authors concluded.
For a lab or bench paper, which controls did they run, and which one is missing?
Why ask it
This is the graduate journal club's version of the bias questions. Take the key figure and ask what each control rules out, then which explanation is still standing. How many times the experiment was repeated, and whether those were separate samples, is the natural second question.
The numbers
What happened in each group, in plain numbers?
Why ask it
Start with the raw counts, before any ratio: how many people in each arm had the outcome, out of how many. Everything else in the results is built from those four numbers, and a room that has them can check the rest for itself. Write them on the board.
What is the absolute difference, and not only the relative one?
Why ask it
Halving a risk sounds the same whether it falls from 20 in 100 to 10 in 100 or from 2 in 10,000 to 1 in 10,000. Subtract one group's rate from the other's, then work out how many people would need the treatment for one of them to benefit. Doing the sum together takes two minutes and often changes the mood of the room.
How wide is the confidence interval, and what would we do at each end of it?
Why ask it
Read the interval as the range of effects the data are reasonably compatible with. If you would act the same way at both ends, the study has answered the question. If one end is a trivial benefit or a harm, it has left the question open, however small the p value.
Is the difference big enough to matter, or only statistically significant?
Why ask it
Before anyone looks at the result, ask the room what the smallest difference worth acting on would be. A very large study can make a tiny difference significant. Check whether the authors named a minimum important difference of their own and whether the result clears it.
How many outcomes and subgroups were tested, and which were planned in advance?
Why ask it
Count them across the tables and the supplement. The more comparisons a paper makes, the more likely one crosses the line by chance. Treat a subgroup finding as a lead for the next study, and ask whether the authors tested for a difference between subgroups or only reported each one.
Is the main result a composite outcome, and which part of it drove the difference?
Why ask it
A composite might combine death, hospital admission and a change in a blood test, and the least serious part is often the most frequent. The components are usually reported one by one in a table or the supplement. If the whole effect sits in the softest one, say so in those words.
What harms were recorded, and how hard did the authors look for them?
Why ask it
Side effects that staff asked about at every visit are counted differently from those patients happened to mention, so the method of collection matters as much as the tally. The number who stopped treatment because of side effects is often the plainest signal. A trial sized to detect a benefit can be too small to see an uncommon harm, so none found is weak reassurance.
Which statistical test did they use, and can someone explain why it fits this data?
Why ask it
Put this one to the room, not only the presenter, and admit freely that you do not know. Often nobody does. If so, write it down for the faculty lead or a statistician to answer afterward, because ten minutes of group guessing teaches nothing.
Does the main figure show what the text says it shows?
Why ask it
With the figure on screen, read the axes first: where the scale starts, what the units are, how many people remain at each time point. Survival curves that separate only at the far right, where few participants are left, carry less weight than they appear to.
Does the conclusion in the abstract match the primary result?
Why ask it
Read the last sentence of the abstract aloud, then the primary outcome line from the results, and ask whether one follows from the other. A conclusion that leans on a secondary outcome or a subgroup when the main result was neutral is an overstatement worth checking for in every paper.
Applying it
Are the people in this study like the people we see?
Why ask it
Set the baseline table beside your own ward, clinic or caseload: age, severity, other conditions, the setting of care. Then name the patients you see most who would have been excluded. For them the paper offers a guess, and the room should say that plainly.
What did this study do well?
Why ask it
Put it to the room before the list of flaws gets long. A large sample, an outcome nobody can argue with, near-complete follow-up or openly shared data are strengths a group skips past once the criticism starts. Saying them aloud keeps the final verdict fair to the paper.
Which limitation matters most for how we would use this?
Why ask it
The authors list their own limitations near the end of the discussion. Compare that paragraph with the list the room has built over the session. What the authors left out is often the most useful thing to talk about, and ranking the flaws stops the group from treating every one as fatal.
Could we deliver this the way the study did?
Why ask it
Trial sites may have had extra staff, closer monitoring or operators who had done the procedure many times. List the training, equipment and follow-up the intervention needed. Whether that exists where you work is a local question, so put it to someone who knows your unit.
How does this fit with the other evidence on the question?
Why ask it
One study seldom settles a question. Ask whether anyone knows of earlier trials, a systematic review or a current guideline, and whether that guideline was written before this paper came out. If the result stands alone against the rest, ask what is different about this study.
What would it cost in money, time and burden on the patient?
Why ask it
Count the extra visits, blood tests and monitoring as well as the price. Prices and coverage differ by country, health system and insurer, so nobody should quote a figure from memory. Ask the pharmacist or whoever handles purchasing how it works where you are.
How would we explain this result to a patient in one sentence?
Why ask it
Have two or three people try it aloud, using absolute numbers and no jargon. If nobody can, the room has not finished understanding the paper. It is good practice for students, and it shows quickly whether the benefit sounds worth the trouble.
Would this change what we do on Monday morning?
Why ask it
Ask each person for yes, no or not yet, with one reason. A split room is where the best discussion starts. An actual change in practice goes through whoever sets protocols and guidelines in your workplace, so ask how that process runs there before acting on a single paper.
Is there a method in this paper we should borrow for our own projects?
Why ask it
In a graduate journal club the takeaway is usually a technique: a control you are not running, an analysis worth copying, a measure worth switching to. Have each person name one thing they would do differently in their own project, even a small one.
What study would we want to see next?
Why ask it
Naming the next study forces the room to say exactly what is still unknown: a longer follow-up, a different population, a head-to-head comparison. Check a trial registry to see whether someone is already running it. For a student, the answer is sometimes a project idea.
Presenter and room
Why did you choose this paper?
Why ask it
A friendly opener that usually uncovers the case or the argument that sent the presenter looking. That story gives everyone a reason to care about the result. If a supervisor assigned the paper, ask the supervisor instead.
Can we go back to that table for a moment?
Why ask it
Possibly the most useful sentence a quiet attendee can say. Name the table or figure and the row you are looking at, so the whole room is on the same number. Presenters often move fastest through the slides they are least sure of, and this slows things down without criticizing anyone.
Can someone explain what that term means?
Why ask it
Ask it the first time you are lost, not the third. If you do not know what a hazard ratio or a non-inferiority margin is, several others likely do not either, and a senior person asking does more for the room than a lecture would. Addressing it to 'someone' keeps the presenter off the spot.
Which part of the paper did you find hardest to follow?
Why ask it
It gives the presenter permission to admit a gap, and turns that gap into something the group works out together. It lands best coming from a senior person. Follow it by opening the paper to that section, not by explaining it from memory.
Did you look at the supplement or the protocol, and was anything there that the main paper left out?
Why ask it
Only ask if 'no' is an acceptable answer, since many presenters have not had time. The supplement often holds the full outcome list, protocol changes and extra analyses. Offering to pull it up together keeps the question from sounding like a test.
Has anyone written a response to this paper, and what did they say?
Why ask it
Editorials and letters to the journal can raise the strongest criticism in a few paragraphs, and the authors' reply shows how well it holds up. It is also worth checking the journal's page for a correction or retraction. If the presenter has not looked, someone with a phone can in under a minute.
If you had reviewed this for the journal, what would you have asked the authors to change?
Why ask it
This turns a summary into an appraisal without making the presenter defend the paper. Good answers are specific: a missing analysis, a table that should exist, a claim to tone down. Afterward, ask the room whether the change would have altered the conclusion.
What does someone who works with these patients every day make of this?
Why ask it
A way to bring in the senior clinician, the nurse or the pharmacist who has said nothing yet, and a way for a trainee to hand a question upward. Experience is not evidence, but it is good at spotting where a trial's tidy conditions part from the ward.
What is your own bottom line: do you believe it?
Why ask it
Presenters often summarize a paper without committing to a view. Ask near the end, kindly, and follow with what would change their mind. Then put the same question to the room, so the presenter is not the only one on the record.
What is the one thing each of us is taking away?
Why ask it
Go around the room, one sentence each, and have someone write them down. Where the answers differ widely, the session left something unclear, and the organizer knows what to pick up next time. It also ends the hour on a conclusion instead of the clock.
How to ask questions in journal club
Practical guidance for the conversation itself
Before the session
Read the methods and tables first
The abstract tells you what the authors want you to conclude. Read the methods, the flow diagram and the first two tables before the discussion, and write down what you think the study showed. Then see whether the authors agree with you. The places where they do not are your questions.
Work out what kind of paper it is
The Design and bias group is written with trials in mind. Other study types has one question each for observational studies, studies of a test, reviews, qualitative papers and lab papers, and only one of them will fit the paper in front of you. Identify the design from the methods, then cross out what does not apply. Some programs hand out a critical appraisal checklist matched to the design, so ask your organizer whether yours uses one and bring it along.
Bring two questions, written down
One question about the methods and one about applying the result is enough for most sessions. Beside each, note the page, table or figure it refers to. A question tied to a line in the paper is easy to say aloud, and having a second means you still have something when another attendee asks your first.
Find out how your journal club runs
Formats differ between programs. Some take questions throughout, some hold them until the presenter has finished, and some assign a discussant or split the room into small groups. Ask the organizer or a colleague who has been before, and ask whether attendance or participation is assessed.
Raising a question without derailing the session
Point at the page
'In Table 2, the third row, is that the primary outcome?' gets a quick answer and brings everyone to the same number. 'I was not convinced by the results' starts a speech. Give the location first, then the question.
Question the paper, not the presenter
The presenter did not run the study. 'Do the authors say how they handled dropouts?' invites them to look with you, while 'Why did you not mention the dropouts?' puts them on trial. The same doubt, phrased about the authors, keeps a junior presenter willing to volunteer again.
Match the question to the moment
A clarifying question, such as what a term means or which group a number belongs to, helps most at the slide where the confusion starts. Larger questions about bias or whether the result applies usually belong after the results have been shown. If the chair has set a different rule, follow it.
One question, then one follow-up
Ask, listen and follow up once. If the point is still open, say 'maybe we can come back to that' and let someone else in. Three questions in a row from one person turn a discussion into an oral exam, even when every one of them is good.
Say when you do not know
'I do not understand how they got from this number to that one' is a real contribution, and usually a relief to half the room. Journal club exists for practicing how to read a paper, so a question that exposes a gap is the format working as intended.
If you are the one presenting
Pick three questions to put to the room
Choose one from Design and bias, one from The numbers and one from Applying it, and decide which slide each belongs on. Stop at that slide and ask before you give your own view. A room that has been asked something in the first ten minutes tends to keep talking.
Appraise the paper yourself first
Work through the list alone and mark where the paper is weak. You do not have to present every flaw. Hold one or two back, and if nobody raises them, try 'did anyone notice anything about who was excluded?' in place of announcing it.
When you cannot answer
Say so, and turn it outward: 'I do not know. Does anyone here?' If no one does, write the question down and offer to send the answer round afterward. Guessing in front of a statistician or a specialist costs more than the gap would have.
Leave time to discuss
The familiar failure is a careful forty-minute summary followed by two minutes of questions. Ask your organizer how long you have and plan to finish presenting with about a third of it left. Cut background slides before you cut the discussion.
Where a journal club goes wrong
Hunting flaws for sport
Every study has limitations, and a room can spend an hour listing them and learn nothing. For each flaw, ask which direction it would push the result, and whether by enough to change the conclusion. A bias that would have hidden a benefit is no reason to doubt a benefit that showed up anyway. Asking what the study did well, early on, is the simplest guard against this.
The statistics detour
One unfamiliar method can swallow the session. If nobody present can explain it within a couple of minutes, park it, name who will find out, and go back to what the result means. The absolute difference and the confidence interval can nearly always be discussed without it.
Two voices and a silent room
When the same two senior people are trading points, the chair or any attendee can redirect with a question aimed elsewhere: 'What do the students think?' or 'How does this look from pharmacy?' The questions under Presenter and room that are addressed to 'someone' or 'each of us' are there for this.
Ending without a verdict
A session that stops when the clock does leaves people remembering the slides and little else. Keep five minutes for the Monday morning question and the one-sentence takeaway. Even 'we do not think this should change anything, and here is why' is a result worth recording.
Treating one paper as the answer
Journal club is for learning to read evidence, and it does not set policy. If the group thinks a paper should change practice, the next step is to take it to whoever is responsible for guidelines or protocols where you work and ask how a change gets proposed there.