Contents
On this page
- Elements that the primary question of a cohort study must specify
- The need for sub-questions in a long-term cohort study
- Formulating sub-questions anchored to the primary question
- A worked example: a primary question and three sub-questions
- Linking each question to a data-collection plan
- Registration of the question and sub-questions before analysis
- Revision of a research question during the study
Academy · Research Data, Analytics & AI · Framing Cohort Study Questions
Framing Research Questions and Sub-Research Questions for Prospective Cohort Studies
The primary research question of a prospective cohort study must specify an exposure, an outcome and a follow-up period before data collection begins. Division of the primary question into clearly scoped sub-questions prevents a long follow-up period from diverging from what the study was designed to measure.
A prospective cohort study follows a defined, currently exposure-free or exposure-varying group forward in time to determine who develops an outcome, which means that its research question must specify three elements before data collection starts: the exposure being tracked, the outcome being measured, and the follow-up period over which that outcome will be assessed. Practical guidance on structuring research questions, developed originally for surgical research but widely applied across clinical and epidemiological study design, frames this specification step as the point at which a vague clinical idea becomes a fundable, answerable study design, and the same discipline prevents a multi-year cohort study from losing focus during follow-up.
Learning objectives
- The elements that the primary research question of a cohort study must specify before data collection begins
- Why and how a primary question should be divided into sub-questions in a long-running cohort study
- A worked example showing a primary question and three supporting sub-questions
Common misconceptions
- “A broad, open-ended question like 'what factors affect long-term health outcomes in this population' is a reasonable starting point for a prospective cohort study.” Widely used guidance on structuring research questions for clinical and epidemiological studies holds that a properly framed question specifies the exposure, the population, the outcome, and the timeframe explicitly before a study begins, precisely because an open-ended question of this kind cannot specify in advance which data to collect, how often, or for how long, decisions that a prospective design must make before recruitment starts.
Elements that the primary question of a cohort study must specify
A prospective cohort study’s primary research question should name the exposure being tracked, the population within which it is being tracked, and the outcome that will eventually be measured, together with the follow-up period over which that outcome will be assessed. Farrugia and colleagues’ widely used guidance on framing research questions for clinical studies treats this specification as a prerequisite for writing a testable hypothesis at all, arguing that an insufficiently specific question cannot be converted into a clear, falsifiable prediction, which in turn cannot be matched to an appropriate statistical test1. For a prospective design specifically, the follow-up period is not a minor detail; it determines sample-size calculations, the budget and duration of the study, and which statistical approach, binary-outcome or time-to-event, the eventual analysis will use2.
The need for sub-questions in a long-term cohort study
A prospective cohort study often runs for months or years, during which multiple secondary outcomes, subgroup patterns, or time-varying effects can reasonably be examined without diluting the study’s primary purpose, provided that these are explicitly planned as sub-questions rather than left to emerge informally from the data as follow-up proceeds. The STROBE reporting guidelines for observational studies specifically call for authors to distinguish primary from secondary objectives clearly in the methods section, which is only possible if that distinction was built into the research question structure from the start rather than retrofitted once results are available3. Sub-questions planned in advance also protect against a specific analytic risk: testing many unplanned associations within a large cohort dataset increases the chance of reporting a spurious, statistically significant finding that does not reflect a real effect, a risk that a pre-specified, limited set of sub-questions is designed to contain4.
Formulating sub-questions anchored to the primary question
Each sub-question should share the same exposure or the same outcome as the primary question, changing only one additional dimension, a subgroup, a secondary outcome, a different follow-up timepoint, or a candidate moderator, rather than introducing an entirely separate research idea under the same study. At this point the distinction between independent, dependent, and moderator variables becomes directly relevant: a sub-question asking whether the primary exposure-outcome relationship differs by a specific subgroup is testing a moderator and requires its own, explicitly planned interaction analysis, not merely an additional subgroup table added retrospectively5. Writing sub-questions this way, as variations on a shared core rather than independent questions, keeps the study's overall data-collection plan coherent and keeps the final set of analyses traceable back to a plan that existed before the data did.
A worked example: a primary question and three sub-questions
In a hypothetical cohort of adults newly diagnosed with hypertension, followed for five years, the primary question might be: “Among adults newly diagnosed with hypertension, does low baseline physical activity increase the five-year risk of a major cardiovascular event, compared with moderate or high baseline physical activity?” A first sub-question retains the same exposure and outcome but adds a temporal dimension: “Does this association differ between the first two years of follow-up and years three through five?” A second sub-question introduces a candidate moderator: “Does the association between baseline physical activity and cardiovascular risk differ by baseline body mass index category?” A third sub-question changes the outcome while retaining the exposure and population: “Does baseline physical activity predict five-year all-cause mortality in the same cohort?” Each sub-question reuses the core structure of the primary question, which keeps the overall study plan, and its data-collection requirements, coherent across a multi-year follow-up period.
Linking each question to a data-collection plan
Once the primary question and its sub-questions are written, each one should be checked against the study’s actual data-collection schedule before recruitment begins, since a sub-question that requires a measurement the protocol does not plan to collect is not answerable regardless of how well it is worded. For the hypertension cohort example, the body-mass-index moderator sub-question requires BMI to be measured and recorded at baseline with sufficient precision to support category-based subgroup analysis, not simply noted informally in a clinical record. The time-varying sub-question requires the cardiovascular-event outcome to be captured with sufficient date precision to distinguish early from late events, which in turn requires a follow-up visit schedule frequent enough to detect events close to the time they occur, rather than only at the final five-year assessment. The all-cause-mortality sub-question requires a reliable mechanism for tracking deaths among participants who may otherwise be lost to routine follow-up contact, such as linkage to a national death registry. Reviewing each sub-question in this way, establishing the data point, the precision and the schedule on which it depends, converts a well-written set of questions into a data-collection plan that a study can execute.
Registration of the question and sub-questions before analysis
Writing a precise primary question and a pre-specified set of sub-questions is most valuable when that exact set is also registered or otherwise time-stamped before the cohort’s data are analyzed, since a registered plan later allows a reader to distinguish a genuinely pre-specified finding from one discovered by exploring the dataset retrospectively and then reported as though it had been anticipated from the outset. For a prospective cohort study, this can take the form of a detailed analysis plan filed with a funder, an institutional protocol registered before enrollment closes, or, where appropriate, a public study registry entry, and should include not only the primary question and sub-questions but also the specific statistical model each one will use, so that readers can later check whether the published analysis matches what was planned. This additional step is not a bureaucratic formality; it is the mechanism by which a reader can distinguish a cohort study that tested a small, disciplined set of pre-specified questions from one that ran dozens of undisclosed, unplanned comparisons until a few reached statistical significance.
Revision of a research question during the study
Occasionally a genuine need to revise the primary question or add a sub-question arises partway through a multi-year cohort study, typically because new external evidence changes which outcome is most clinically relevant, or because an unanticipated data-quality problem makes a planned measure unusable. In such cases, the appropriate approach is to document the change explicitly in a date-stamped protocol amendment, recording what the original question was, what it was changed to, and why, rather than revising the write-up later without disclosure as though the final version had been the plan from the outset. A transparent amendment record weakens a study’s credibility far less than an undisclosed, retrospectively concealed change in direction would if a careful reader later identified the inconsistency between an early protocol and the final published question. Funders and institutional review boards reviewing a protocol amendment also tend to regard a well-justified, clearly documented change far more favorably than an unexplained one discovered independently, since the former demonstrates that the research team is actively managing the study’s scientific integrity rather than allowing it to drift unnoticed or reacting only once an external reviewer raises the discrepancy.
Key points
Key points
The primary question of a prospective cohort study must specify the exposure, population, outcome and follow-up period before data collection begins; any sub-questions should be planned in advance as deliberate variations on the same structure, not as an informal and expanding list of analyses once the data exist.
Practical application
Practical application
- The primary question should state the exposure, population, outcome and follow-up period explicitly. Each of these four elements has direct, practical consequences for sample size, budget, and the eventual statistical analysis.
- Sub-questions should be planned in advance, each varying only one dimension of the primary question. This approach keeps the sub-questions traceable to a pre-specified plan and limits the risk of reporting a spurious, data-driven association as if it had been planned.
- The additional dimension of each sub-question should be classified as a subgroup, a secondary outcome, a timepoint or a moderator before data collection. This classification determines which statistical analysis — an interaction term, a time-stratified model or a separate outcome model — the sub-question will require.
Frequently asked questions
There is no fixed limit, but each additional sub-question increases the number of statistical tests run against the same dataset, so the number should be justified by genuine prior interest or a specific biological or clinical rationale, not simply by the number of associations the available data permit.
Yes, wherever possible; pre-specifying sub-questions in a trial or study registry or a published protocol, before the data are collected or at least before they are analyzed, is one of the clearest ways to demonstrate that a reported association was planned rather than discovered after the fact.
It can still be reported, but should be explicitly labeled as an exploratory or post-hoc finding rather than presented alongside the pre-specified primary and sub-questions as if it had equal evidentiary weight; STROBE-style reporting expects this distinction to be stated clearly.
Ideally, yes, at least for sub-questions involving a subgroup analysis or a moderator test, since the subgroup or interaction-effect sample size needed to detect a real effect is typically larger than the sample size needed for the primary question's main effect alone.
References
- Farrugia P, Petrisor BA, Farrokhyar F, Bhandari M. Practical tips for surgical research: Research questions, hypotheses and objectives. Canadian Journal of Surgery. 2010;53(4):278-281. read 10 October 2026. View on PMC
- Song JW, Chung KC. Observational Studies: Cohort and Case-Control Studies. Plastic and Reconstructive Surgery. 2010;126(6):2234-2242. read 10 October 2026. View on PMC
- von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP; STROBE Initiative. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Lancet. 2007;370(9596):1453-1457. read 10 October 2026. View on The Lancet
- Sterne JAC, Davey Smith G. Sifting the evidence-what's wrong with significance tests? BMJ. 2001;322(7280):226-231. read 10 October 2026. View on BMJ
- Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology. 1986;51(6):1173-1182. read 10 October 2026. View via DOI
- Cox DR. Regression models and life-tables. Journal of the Royal Statistical Society Series B. 1972;34(2):187-220. read 10 October 2026. View via DOI
- Richardson WS, Wilson MC, Nishikawa J, Hayward RS. The well-built clinical question: a key to evidence-based decisions. ACP Journal Club. 1995;123(3):A12-A13. read 10 October 2026. View via DOI
- Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71. read 10 October 2026. View on BMJ
- Munn Z, Moola S, Riitano D, Lisy K. The development of a critical appraisal tool for use in systematic reviews addressing questions of prevalence. International Journal of Health Policy and Management. 2014;3(3):123-128. read 10 October 2026. View on IJHPM
- International Committee of Medical Journal Editors. Recommendations for the Conduct, Reporting, Editing, and Publication of Scholarly Work in Medical Journals. read 10 October 2026. View ICMJE Recommendations
Every source above is a peer-reviewed article, a publisher’s record of one, or a standard issued by the body that owns it. Each carries the date it was last read. Where a standard moves, this article is revised rather than left up with a note.
How to cite this article
Pick your referencing style and copy it. Your university or target journal may vary the punctuation — check its guide if it has one.
Pubrica Academy. (2026, October 10). Framing Research Questions and Sub-Research Questions for Prospective Cohort Studies. Pubrica. https://pubrica.com/academy/data-collection/how-to-frame-a-research-question-and-sub-research-question-for-prospective-cohort-studies/
Pubrica Academy. “Framing Research Questions and Sub-Research Questions for Prospective Cohort Studies.” Pubrica, 10 Oct. 2026, pubrica.com/academy/data-collection/how-to-frame-a-research-question-and-sub-research-question-for-prospective-cohort-studies/.
Pubrica Academy (2026) Framing Research Questions and Sub-Research Questions for Prospective Cohort Studies. Pubrica. Available at: https://pubrica.com/academy/data-collection/how-to-frame-a-research-question-and-sub-research-question-for-prospective-cohort-studies/ (Accessed: 15 October 2026).
Pubrica Academy. “Framing Research Questions and Sub-Research Questions for Prospective Cohort Studies.” Pubrica. October 10, 2026. https://pubrica.com/academy/data-collection/how-to-frame-a-research-question-and-sub-research-question-for-prospective-cohort-studies/
Pubrica Academy, ‘Framing Research Questions and Sub-Research Questions for Prospective Cohort Studies’ (Pubrica, 10 October 2026) <https://pubrica.com/academy/data-collection/how-to-frame-a-research-question-and-sub-research-question-for-prospective-cohort-studies/> accessed 15 October 2026.
Pubrica Academy. Framing Research Questions and Sub-Research Questions for Prospective Cohort Studies. [Internet]. Pubrica; 2026 Oct 10 [cited 2026 Oct 15]. Available from: https://pubrica.com/academy/data-collection/how-to-frame-a-research-question-and-sub-research-question-for-prospective-cohort-studies/
Pubrica Academy, “Framing Research Questions and Sub-Research Questions for Prospective Cohort Studies,” Pubrica, Oct. 10, 2026. [Online]. Available: https://pubrica.com/academy/data-collection/how-to-frame-a-research-question-and-sub-research-question-for-prospective-cohort-studies/
@misc{/la_framing_research_question_cohort_studies_2026,
author = {{Pubrica Academy}},
title = {Framing Research Questions and Sub-Research Questions for Prospective Cohort Studies},
howpublished = {Pubrica},
year = {2026},
month = {oct},
url = {https://pubrica.com/academy/data-collection/how-to-frame-a-research-question-and-sub-research-question-for-prospective-cohort-studies/},
note = {Accessed: <span data-cite-acc="iso">2026-10-15</span>}
}This article is written to be learned from. Cite it if you are referring to what it says; if you are using the evidence behind it, cite the sources listed above instead.
Cohort Study Question Worksheet
This template is intended for framing a primary research question and its sub-questions before data collection begins.
Primary question: Exposure: ____________________ Population: ____________________ Outcome: ____________________ Follow-up period: ____________________ Sub-question 1 (varies: subgroup / secondary outcome / timepoint / moderator - circle one): ____________________ Sub-question 2 (varies: ______): ____________________ Sub-question 3 (varies: ______): ____________________ [ ] Confirmed each sub-question changes only one dimension from the primary question [ ] Classified each sub-question's analysis type (interaction, time-stratified, separate outcome model) [ ] Pre-specified all sub-questions in the protocol before data collection or analysis begins
Pubrica AcademyWritten and maintained by Pubrica’s subject matter experts. Articles are revised when the underlying standards change, and each article states the date on which it was last reviewed.
How Pubrica can help
Pubrica’s research design team frames the primary research questions and sub-questions of prospective cohort studies before data collection, specifying the exposure, population, outcome and follow-up period explicitly and matching each sub-question to the analysis it requires.