Practical guidePublication & ethicsMedicine & healthcare

STROBE: reporting an observational study without randomisation

Cohort, case-control and cross-sectional studies face a different reporting problem from trials: confounding, not allocation. This guide explains what STROBE asks you to report, which observational designs it covers, where it stops, and how to write a methods section that lets a reader judge the risk of confounding for themselves.

What you will learn

  • Explain what STROBE is and the reporting problem specific to observational studies
  • Identify which of the three main observational designs (cohort, case-control, cross-sectional) your study is
  • Report how confounders and other variables were measured, defined and handled
  • Recognise where STROBE should give way to a trial, diagnostic or systematic review guideline instead
  • Compare a thin observational methods section against one written to STROBE

What you will unlearn

  • Any health-related study that is not a trial should use STROBE.STROBE covers cohort, case-control and cross-sectional designs specifically; diagnostic accuracy studies use STARD and systematic reviews use PRISMA.
  • Because there is no randomisation, confounding cannot really be addressed, only mentioned.STROBE expects explicit reporting of how confounders were identified, measured and adjusted for, and how residual confounding might affect the result.
  • STROBE is only relevant to epidemiology.It applies to observational designs across health, social and biological research wherever exposure and outcome are measured without intervention by the researcher.
  • A large sample size compensates for weak control of confounding.A larger sample narrows the confidence interval around a biased estimate; it does not remove the bias itself.

Skills you will practise

  • Naming the observational design precisely
  • Reporting confounders and adjustment
  • Describing participant selection
  • Handling missing data in observational studies
  • Discussing bias honestly

What STROBE is and who maintains it

STROBE, the Strengthening the Reporting of Observational Studies in Epidemiology statement, is a 22-item checklist for reporting cohort, case-control and cross-sectional studies. It was developed by an international collaboration of methodologists, epidemiologists, statisticians and journal editors and first published in 2007 (von Elm et al., 2007).

The checklist covers the title and abstract, introduction, methods (design, setting, participants, variables, data sources, bias, study size, statistical methods), results and discussion. Several items apply, with minor variations, across all three designs; a small number are specific to one design, such as how cases and controls were matched in a case-control study.

STROBE sits alongside general statistical reporting guidance, such as the SAMPL guidelines for basic statistical reporting in biomedical journals (Lang & Altman, 2015); SAMPL tells you how to report a given estimate or test correctly, while STROBE tells you which design features, variables and sources of bias an observational study report has to disclose in the first place.

Why it exists

An observational study cannot use randomisation to balance known and unknown factors between exposure groups, so the association it reports between exposure and outcome may partly reflect confounding rather than a genuine effect. Readers can only judge how much confounding is a concern if the report tells them which confounders were considered, how they were measured, and how they were handled in the analysis. Reports that state only that "adjustment was made for confounders" without naming them give a reader nothing to evaluate.

Selection of participants raises a related, separate problem. Who was eligible, who was approached, and who actually took part can each introduce bias that has nothing to do with confounding by a measured variable; a cross-sectional survey with a 20% response rate, for instance, may differ systematically from the population it meant to describe in ways no amount of statistical adjustment can fix. STROBE asks for these participant flow numbers precisely so a reader can weigh this risk.

STROBE exists because this problem was, and remains, common: a study can be well conducted but poorly reported, leaving a reader unable to tell the difference between a carefully controlled analysis and a superficial one. The checklist gives authors, reviewers and readers a shared, explicit list of what an observational study report needs to say. STROBE is one of the guidelines coordinated through the EQUATOR Network's wider library of reporting standards for health research (Moher et al., 2008).

Which designs it fits

STROBE is written for the three classical observational designs, and the core 22-item checklist applies to all of them, with design-specific notes for a handful of items.

STROBE-covered designs and what to report for each
DesignWhat it measuresSTROBE-specific reporting point
Cohort studyExposure followed forward to outcome over timeReport the method of follow-up and how loss to follow-up was handled
Case-control studyCases and controls compared backward for prior exposureReport how cases and controls were defined and selected, and any matching criteria
Cross-sectional studyExposure and outcome measured at a single point in timeMake clear that temporal (cause-and-effect) order cannot be established from the design alone

STROBE also has extensions for specific applications, such as reporting genetic association studies (STREGA) and molecular epidemiology, which add items relevant to those fields without replacing the core checklist. A doctoral study that combines, say, a cohort design with a genetic sub-study should use the core checklist for the main design and check whether the relevant extension adds items for the genetic component, rather than treating one or the other as sufficient on its own.

Where STROBE does not apply

STROBE is not a substitute for CONSORT when a study involves any form of researcher-controlled allocation to groups, even an imperfectly randomised or quasi-experimental one; such studies should be reported, and ideally designed, with CONSORT or its non-randomised study extensions in mind, since the items that matter most, such as allocation and blinding, differ sharply from those in STROBE (Hopewell et al., 2025).

STROBE also does not cover diagnostic test accuracy studies, which have their own concerns around index tests, reference standards and patient spectrum, nor does it cover the reporting of a systematic review of observational studies, which should instead follow PRISMA while still drawing on STROBE items to judge the quality of the individual studies included (Page et al., 2021). Case reports and small case series describing a single patient or a handful of patients are better reported with CARE, since STROBE's items on sampling and confounder adjustment assume a study population large enough for comparison.

Why it matters

Observational studies make up a large share of the evidence base in fields where trials are impractical or unethical, including most of environmental, occupational and much clinical epidemiology. The certainty that can be placed in this evidence depends heavily on how well confounding, selection and measurement bias have been handled and disclosed; rating systems such as GRADE explicitly take the risk of bias in observational evidence into account when judging how much confidence to place in a conclusion drawn from it (Guyatt et al., 2008). Reporting that hides these details, even unintentionally, makes that judgement harder to make well.

How it is applied in practice

Many epidemiology, public health and clinical journals require a completed STROBE checklist at submission, cross-referenced to page numbers in the manuscript, in the same way many journals require a CONSORT checklist for trials. STROBE is also widely used as a teaching and critical appraisal tool: supervisors and reviewers often use its items as a checklist when reading a draft observational methods section, independent of where it will eventually be submitted.

Systematic reviewers also use STROBE items, informally or through a related risk-of-bias tool, to appraise the observational studies they include; a study that reports its confounders and selection process clearly is easier to rate for risk of bias than one that does not, which gives authors a further, indirect incentive to report fully even when no journal checklist is enforced at submission.

In a doctoral thesis, applying STROBE early, while the study is still being designed, is more useful than applying it only when writing up, because several of its items, such as how a confounder will be measured and when, are easier to get right before data collection than to repair afterwards.

A supervisor reviewing a draft chapter against STROBE item by item can usually spot missing information quickly: a methods section that does not say how a key confounder was defined, or a results section that gives adjusted estimates without saying what they were adjusted for, stands out as soon as the checklist is applied, even to a reader unfamiliar with the specific study.

An illustrative example

The passage below is a hypothetical illustration, not a quotation from a published study.

Illustrative before-and-after reporting against STROBE (hypothetical cohort study)
ElementWithout STROBEWith STROBE
Design"A study was carried out on diet and diabetes.""A prospective cohort study followed 2,340 adults without diabetes at baseline for a median of 6.2 years to examine the association between dietary pattern and incident type 2 diabetes."
Participants"Participants were recruited from clinics.""Participants were recruited from four primary care clinics between January 2020 and June 2021; eligibility and exclusion criteria are listed, and the number approached, eligible and enrolled is reported."
Confounders"Results were adjusted for confounders.""Models were adjusted for age, sex, baseline BMI, physical activity and family history of diabetes, each measured as described in Table 1; the rationale for including each variable is given."
Missing dataNot stated"8% of participants had missing dietary data; these were excluded from the primary analysis, and a sensitivity analysis using multiple imputation produced similar estimates."
Limitations"The study has some limitations.""Residual confounding by unmeasured factors such as detailed physical activity cannot be excluded; the cross-sectional measurement of diet at baseline may not reflect long-term intake."

What to check before you submit

  • Does the title or abstract name the specific design (cohort, case-control or cross-sectional)?
  • Are eligibility criteria, and the numbers approached, eligible, and analysed, reported?
  • Are all confounders and other key variables named, with how and when each was measured?
  • Is the statistical method for handling confounding, and any sensitivity analysis, described?
  • Is missing data reported with its extent and how it was handled?
  • Does the discussion address likely sources of bias, including residual confounding, honestly?
  • Official checklist and source: STROBE Statement (strobe-statement.org) .

Key takeaways

  • STROBE is a 22-item checklist for reporting cohort, case-control and cross-sectional studies (von Elm et al., 2007).
  • It exists because, without randomisation, a reader can only judge confounding if the report names and describes it.
  • It fits the three classical observational designs and their STREGA-type extensions, not trials (Hopewell et al., 2025) or systematic reviews (Page et al., 2021).
  • How well confounding and bias are reported feeds directly into how observational evidence is later rated, for example with GRADE (Guyatt et al., 2008).
  • Applying STROBE items while designing the study, not only while writing it up, makes several items far easier to satisfy.
  • STROBE complements, rather than replaces, general statistical reporting guidance such as SAMPL (Lang & Altman, 2015).
Your next step

Want to apply this to your own study?

  1. 1Tell us your topic, data and deadline
  2. 2A subject mentor reviews your work with you
  3. 3You get a plan and a quote
Manuscript statistics

References

  1. Guyatt, G. H., Oxman, A. D., Vist, G. E., Kunz, R., Falck-Ytter, Y., Alonso-Coello, P., & Schünemann, H. J. (2008). GRADE: An emerging consensus on rating quality of evidence and strength of recommendations. BMJ, 336(7650), 924–926. https://doi.org/10.1136/bmj.39489.470347.AD
  2. Hopewell, S., Chan, A.-W., Collins, G. S., Hróbjartsson, A., Moher, D., Schulz, K. F., Tunn, R., Aggarwal, R., Berkwits, M., Berlin, J. A., Bhandari, N., Butcher, N. J., Campbell, M. K., Chidebe, R. C. W., Elbourne, D., Farmer, A., Fergusson, D. A., Golub, R. M., Goodman, S. N., . . . Boutron, I. (2025). CONSORT 2025 statement: Updated guideline for reporting randomised trials. BMJ, 389, e081123. https://doi.org/10.1136/bmj-2024-081123
  3. Lang, T. A., & Altman, D. G. (2015). Basic statistical reporting for articles published in biomedical journals: The "Statistical Analyses and Methods in the Published Literature" or the SAMPL guidelines. International Journal of Nursing Studies, 52(1), 5–9. https://doi.org/10.1016/j.ijnurstu.2014.09.006
  4. Moher, D., Simera, I., Schulz, K. F., Hoey, J., & Altman, D. G. (2008). Helping editors, peer reviewers and authors improve the clarity, completeness and transparency of reporting health research. BMC Medicine, 6, 13. https://doi.org/10.1186/1741-7015-6-13
  5. Page, M. J., McKenzie, J. E., Bossuyt, P. M., Boutron, I., Hoffmann, T. C., Mulrow, C. D., Shamseer, L., Tetzlaff, J. M., Akl, E. A., Brennan, S. E., Chou, R., Glanville, J., Grimshaw, J. M., Hróbjartsson, A., Lalu, M. M., Li, T., Loder, E. W., Mayo-Wilson, E., McDonald, S., . . . Moher, D. (2021). The PRISMA 2020 statement: An updated guideline for reporting systematic reviews. BMJ, 372, Article n71. https://doi.org/10.1136/bmj.n71
  6. von Elm, E., Altman, D. G., Egger, M., Pocock, S. J., Gøtzsche, P. C., & Vandenbroucke, J. P. (2007). The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: Guidelines for reporting observational studies. The Lancet, 370(9596), 1453–1457. https://doi.org/10.1016/S0140-6736(07)61602-X

How to cite this article

Choose your referencing style, then copy. Check your university's or journal's guide for local variations.

PhD Research Lab Academy. (2026, October 8). STROBE: reporting an observational study without randomisation. PhD Research Lab. https://www.phdassistance.com/academy/guidelines/strobe/

PhD Research Lab Academy. “STROBE: Reporting an Observational Study Without Randomisation.” PhD Research Lab, 8 Oct. 2026, www.phdassistance.com/academy/guidelines/strobe. Accessed [access date].

PhD Research Lab Academy (2026) STROBE: reporting an observational study without randomisation. PhD Research Lab. Available at: https://www.phdassistance.com/academy/guidelines/strobe/ (Accessed: [access date]).

PhD Research Lab Academy. “STROBE: Reporting an Observational Study Without Randomisation.” PhD Research Lab. October 8, 2026. https://www.phdassistance.com/academy/guidelines/strobe.

PhD Research Lab Academy, ‘STROBE: Reporting an Observational Study Without Randomisation’ (PhD Research Lab, 8 October 2026) <https://www.phdassistance.com/academy/guidelines/strobe> accessed [access date].

PhD Research Lab Academy. STROBE: reporting an observational study without randomisation [Internet]. PhD Research Lab; 2026 Oct 8 [cited [access date]]. Available from: https://www.phdassistance.com/academy/guidelines/strobe/

PhD Research Lab Academy, “STROBE: reporting an observational study without randomisation,” PhD Research Lab, Oct. 8, 2026. [Online]. Available: https://www.phdassistance.com/academy/guidelines/strobe. [Accessed: [access date]].

@online{guideline_strobe_2026,
  author = {{PhD Research Lab Academy}},
  title = {{STROBE: reporting an observational study without randomisation}},
  organization = {PhD Research Lab},
  date = {2026-10-08},
  url = {https://www.phdassistance.com/academy/guidelines/strobe},
  urldate = {[access date]}
}

This article is for learning. Use it to understand the method and cite the original sources above in your own work; do not copy or submit it as your own.

STROBENeed help applying it to your study?
Manuscript statistics