Contents
Academy · Research Data, Analytics & AI · Data Extraction for Cohort Studies
Variables Used in Data Extraction for Prospective Cohort Studies in a Systematic Review
A data extraction form for prospective cohort studies requires six linked categories: study characteristics, population, exposure, outcomes, results and quality. The form should be pilot-tested before full extraction begins and checked by more than one reviewer.
Data extraction is the structured process of pulling predefined variables out of each study included in a systematic review and recording them in a standardized form, so that results across studies with different designs, populations, and reporting styles can be compared on common terms. For prospective cohort studies, in which participants are followed forward over time from a defined exposure status to a measured outcome, the extraction form must capture not only the final effect estimate but also the structural features (duration of follow-up, loss to follow-up, and confounder adjustment) that determine how much weight that estimate deserves in the synthesis. A form lacking any of these structural fields obliges a reviewer either to return to the original paper mid-analysis or, more commonly, to synthesize results without being able to account properly for differences in study quality.
Learning objectives
- The six linked categories a cohort-study data extraction form needs to cover
- Why pilot-testing the extraction form matters before full extraction begins
- How dual extraction and reviewer consensus reduce errors that a single extractor would miss
Common misconceptions
- “Data extraction is mainly about recording each study's final effect estimate.” Published systematic review guidance treats the effect estimate as only one of several linked extraction categories; without matched data on exposure definition, follow-up duration, loss to follow-up, and confounder adjustment, a reviewer cannot properly judge how much weight that effect estimate deserves relative to other included studies.
Six core categories of a cohort-study extraction form
A complete extraction form for prospective cohort studies generally organizes variables into six linked categories: study characteristics (authors, year, journal, country, funding source, design); population (sample size, age, sex, baseline health status, inclusion and exclusion criteria); exposure or risk factor definition and measurement; outcome definition, measurement method, and follow-up duration; results, meaning crude and adjusted effect estimates with their confidence intervals; and a formal quality or risk-of-bias assessment15. The Cochrane Handbook's chapter on collecting data presents this list as a minimum rather than an exhaustive template, recommending that review teams add fields specific to their research question while retaining the six-category structure, since the omission of any one category tends to become apparent only once synthesis is under way, when it is considerably more costly to correct3. The reporting requirements of PRISMA 2020 expect the completed review to state, in its methods section, which data items were sought for each included study, which means that the category list of the extraction form is not merely an internal working document but a structure that the published review must disclose4.
Required level of detail in exposure and outcome fields
Exposure fields need to capture not only whether a participant was classified as exposed but also how that classification was made (self-report, biomarker, administrative record, or clinician assessment), since studies using different measurement methods for the same nominal exposure are not directly comparable unless that distinction is recorded5. The same logic applies to outcomes: measurement method, follow-up duration, and loss-to-follow-up rate all affect how much confidence a reader can place in a reported incidence or prevalence figure within the exposed and unexposed groups, and a systematic review that extracts only the final effect estimate without these supporting fields cannot distinguish a well-conducted cohort with long follow-up from a shorter cohort with greater attrition that reports a superficially similar figure1. A living systematic review of data extraction tools and methods notes that confounder and covariate-adjustment fields are frequently among the most inconsistently extracted items across review teams, precisely because studies report their adjustment sets in widely varying formats, which makes a standardized, pilot-tested field for recording exactly which confounders were adjusted for in each study's primary model a disproportionately high-value part of the form2.
Piloting the extraction form before full extraction
Standard systematic review methodology calls for the extraction form to be piloted on a small sample of included studies by all review authors before full extraction begins, with the form revised based on where independent extractors disagreed or found a field ambiguous37. The Joanna Briggs Institute's manual for evidence synthesis describes this piloting step as particularly important for prognosis and etiology reviews based on cohort data, since these designs often report results using varying statistical models (hazard ratios, odds ratios, and risk ratios) interchangeably, which the extraction form must capture and flag in a standardized way so that the synthesis stage does not inadvertently combine incompatible effect measures6. A step-by-step guide to conducting a systematic review with simulation data similarly recommends building explicit “not reported” and “unclear” options into every extraction field rather than leaving a blank, since a blank field is ambiguous between “not applicable,” “not reported by the study,” and “not yet extracted,” whereas an explicit code distinguishes all three8.
Dual extraction, consensus and resolution of disagreement
Cochrane Handbook guidance recommends at least two reviewers extracting data independently for key variables, comparing results, and resolving disagreement through discussion or a third reviewer, specifically because single-reviewer extraction has been shown to introduce avoidable errors that a second, independent pass reliably catches3. In practice, many review teams use a hybrid model for cohort-heavy reviews: one reviewer extracts all studies for consistency, while a second reviewer independently extracts a defined subset, often all studies or a random sample, to check agreement, with any discrepancy resolved in a structured consensus meeting held soon after extraction rather than deferred until the synthesis stage, when the context needed to resolve a disagreement is harder to reconstruct7. Newcastle-Ottawa Scale quality assessment, the most widely used tool for rating non-randomized cohort and case-control studies, should be extracted and scored using this same dual-reviewer process, since its star-rating items for selection, comparability, and outcome assessment require the same kind of judgment calls that benefit from independent cross-checking9.
Role and limitations of automation tools
A living systematic review of data extraction tools found a growing set of software options designed to semi-automate extraction, ranging from tools that pre-populate standard fields like sample size or study design from a paper's structured text, to natural-language-processing tools that attempt to flag relevant sentences for a human reviewer to confirm2. The same review found that full automation, extracting a field without any human verification, remains reliable only for a narrow set of highly standardized, structurally predictable fields, such as publication year or journal name, while judgment-dependent fields, confounder adjustment sets, outcome definitions, and quality-assessment scores among them, still require human extraction or at minimum human verification of an automated first pass2. PRISMA 2020 explicitly added reporting items addressing the use of automation tools during the review process, reflecting that this is now common enough practice to need its own disclosure, and a review using semi-automated extraction should state which fields were automated, which tool was used, and how the automated output was verified, rather than presenting the final extraction table as though every field had been extracted identically4. For cohort-study reviews specifically, where confounder handling and outcome measurement detail carry particular weight in judging study quality, treating automation as a time-saving first pass for structural fields, while keeping exposure, outcome, and quality-assessment fields under full human extraction and dual review, is the balance most current guidance supports.
A practical point that applies regardless of how much automation is used is version control over the extraction form itself: when a pilot or a mid-extraction consensus discussion leads to a revised field definition, every study extracted before that revision needs to be re-checked against the new definition, not only studies extracted afterward, since an extraction form that changes mid-project without a corresponding re-check step can leave the completed dataset internally inconsistent in ways that are difficult to detect once synthesis has begun37. Dating and labelling each version of the extraction form, and maintaining a brief changelog describing what changed and why, requires little additional effort and makes such retrospective consistency checks considerably faster when they are needed.
Finally, the extraction form and its completed data should be archived alongside the published review, either as a supplementary file or in a public repository, since PRISMA 2020 and related reporting standards increasingly expect this level of transparency, and a well-archived extraction dataset lets other researchers build directly on the review's work, for example by adding newly published cohort studies to an existing, well-documented dataset rather than starting extraction from scratch for an update4. Such archiving converts a single extraction effort into a reusable resource for the wider research community working on the same question, rather than a one-time internal document discarded once the review is published.
Key points
Key points
A prospective cohort data extraction form requires six linked categories (study characteristics, population, exposure, outcomes, results and quality assessment), piloted on a small sample before full extraction and checked by at least two independent reviewers to detect avoidable errors.
Practical application
Practical application
- The extraction form should address all six categories before extraction begins. An omitted category, particularly confounder adjustment or follow-up duration, is considerably more costly to remedy once synthesis has begun than to plan for at the outset.
- The form should first be piloted on a small sample of included studies. Piloting identifies ambiguous fields and extractor disagreement while corrections remain inexpensive, rather than after full extraction is complete.
- At least two independent extractors should be used for key variables, with a documented consensus process. Cochrane Handbook guidance treats dual extraction as a reliable way to catch errors a single reviewer would miss, and the consensus process should happen soon after extraction, not deferred to synthesis.
Frequently asked questions
Both, where available. Recording crude and adjusted estimates together allows the review to assess how far confounder adjustment changed the result, which is itself informative when judging the contribution of a study to the synthesis.
An explicit “not reported” code should be used rather than a blank field, since a blank field is ambiguous between not applicable, not reported, and not yet extracted.
It is sometimes used for lower-priority fields to save time, but Cochrane Handbook guidance recommends dual, independent extraction for key variables specifically because single-reviewer extraction has been shown to introduce avoidable errors.
It is extracted and scored as part of the same process, using the same dual-reviewer approach, since its selection, comparability, and outcome items require judgment calls that benefit from independent cross-checking.
Yes. PRISMA 2020 expects the published review's methods section to state which data items were sought for each included study, which makes the extraction form's category structure part of what the review must report, not only an internal working document.
References
- Louie JCY, Flood VM, Hector DJ, Rangan AM, Gill TP. Dairy consumption and overweight and obesity: a systematic review of prospective cohort studies. Obesity Reviews. 2011;12(7):e582-e592. read 10 October 2026. View on Obesity Reviews
- Schmidt L, Olorisade BK, McGuinness LA, Higgins JPT, Thomas J. Data extraction methods for systematic review (semi)automation: a living systematic review. F1000Research. 2021;10:401. read 10 October 2026. View on F1000Research
- Li T, Higgins JPT, Deeks JJ, eds. Chapter 5: Collecting data. In: Cochrane Handbook for Systematic Reviews of Interventions. read 10 October 2026. View on Cochrane Training
- Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71. read 10 October 2026. View on The BMJ
- von Elm E, Altman DG, Egger M, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement. The Lancet. 2007;370(9596):1453-1457. read 10 October 2026. View on The Lancet
- Munn Z, Moola S, Lisy K, Riitano D, Tufanaru C. Chapter 7: Systematic reviews of etiology and risk. In: JBI Manual for Evidence Synthesis. read 10 October 2026. View on JBI Global
- Tawfik GM, Dila KAS, Mohamed MYF, et al. A step by step guide for conducting a systematic review and meta-analysis with simulation data. Tropical Medicine and Health. 2019;47:46. read 10 October 2026. View on Tropical Medicine and Health
- Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of observational studies in epidemiology: a proposal for reporting (MOOSE). JAMA. 2000;283(15):2008-2012. read 10 October 2026. View on JAMA
- Wells GA, Shea B, O'Connell D, et al. The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses. Ottawa Hospital Research Institute. read 10 October 2026. View on OHRI.ca
- Moher D, Liberati A, Tetzlaff J, Altman DG, The PRISMA Group. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Medicine. 2009;6(7):e1000097. read 10 October 2026. View on PLOS Medicine
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). Variables Used in Data Extraction for Prospective Cohort Studies in a Systematic Review. Pubrica. https://pubrica.com/academy/systematic-review/variables-used-in-data-extraction-for-prospective-cohort-studies-in-a-systematic-review/
Pubrica Academy. “Variables Used in Data Extraction for Prospective Cohort Studies in a Systematic Review.” Pubrica, 10 Oct. 2026, pubrica.com/academy/systematic-review/variables-used-in-data-extraction-for-prospective-cohort-studies-in-a-systematic-review/.
Pubrica Academy (2026) Variables Used in Data Extraction for Prospective Cohort Studies in a Systematic Review. Pubrica. Available at: https://pubrica.com/academy/systematic-review/variables-used-in-data-extraction-for-prospective-cohort-studies-in-a-systematic-review/ (Accessed: 15 October 2026).
Pubrica Academy. “Variables Used in Data Extraction for Prospective Cohort Studies in a Systematic Review.” Pubrica. October 10, 2026. https://pubrica.com/academy/systematic-review/variables-used-in-data-extraction-for-prospective-cohort-studies-in-a-systematic-review/
Pubrica Academy, ‘Variables Used in Data Extraction for Prospective Cohort Studies in a Systematic Review’ (Pubrica, 10 October 2026) <https://pubrica.com/academy/systematic-review/variables-used-in-data-extraction-for-prospective-cohort-studies-in-a-systematic-review/> accessed 15 October 2026.
Pubrica Academy. Variables Used in Data Extraction for Prospective Cohort Studies in a Systematic Review. [Internet]. Pubrica; 2026 Oct 10 [cited 2026 Oct 15]. Available from: https://pubrica.com/academy/systematic-review/variables-used-in-data-extraction-for-prospective-cohort-studies-in-a-systematic-review/
Pubrica Academy, “Variables Used in Data Extraction for Prospective Cohort Studies in a Systematic Review,” Pubrica, Oct. 10, 2026. [Online]. Available: https://pubrica.com/academy/systematic-review/variables-used-in-data-extraction-for-prospective-cohort-studies-in-a-systematic-review/
@misc{/la_data_extraction_variables_cohort_studies_systematic_review_2026,
author = {{Pubrica Academy}},
title = {Variables Used in Data Extraction for Prospective Cohort Studies in a Systematic Review},
howpublished = {Pubrica},
year = {2026},
month = {oct},
url = {https://pubrica.com/academy/systematic-review/variables-used-in-data-extraction-for-prospective-cohort-studies-in-a-systematic-review/},
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 Data Extraction Form Checklist
This checklist should be used to confirm that the extraction form covers all six categories before it is piloted.
[ ] Study characteristics: authors, year, journal, country, funding, design [ ] Population: sample size, age, sex, baseline health, inclusion/exclusion criteria [ ] Exposure: definition, measurement method, categories, confounders considered [ ] Outcome: definition, measurement method, follow-up duration, loss to follow-up [ ] Results: crude and adjusted estimates, confidence intervals, p-values [ ] Quality: Newcastle-Ottawa Scale score or equivalent risk-of-bias assessment [ ] Piloted the form on a small sample with at least two independent extractors [ ] Added explicit “not reported” / “unclear” codes to every field
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 systematic review team builds and pilots cohort-study data extraction forms across all six required categories, with dual-reviewer extraction and documented consensus, so that gaps are identified before the synthesis stage.