What is the difference between inclusion and exclusion criteria?

Inclusion criteria are the characteristics a source must have to be considered. Exclusion criteria are characteristics that disqualify a source that would otherwise be included. “Adults aged 18 or over” is an inclusion criterion. “Studies in which more than half the sample were inpatients” is an exclusion criterion. Writing “under 18” as an exclusion criterion adds nothing, because it only repeats the inclusion criterion in negative form.

Do I need inclusion and exclusion criteria for a narrative literature review?

They are not mandatory, but stating a few lets readers see how sources were chosen and makes the review easier to manage. Three or four sentences on publication years, source types, populations and languages covered are enough for a thesis chapter. They also help you say no to interesting papers that do not belong.

Can I limit my review to the last five or ten years?

Only if you can justify it from the topic: a technology that did not exist before a certain year, a policy change, a revised diagnostic definition, or an earlier review that already covers the period before. The Cochrane Handbook says that applying arbitrary date limits is not recommended for Cochrane reviews of interventions, and gives as a justified limit the approval date of the first device of the kind under review. “To keep the numbers manageable” is an arbitrary limit in that sense. For coursework, ask your supervisor, and state the limit openly if you use one.

Is it acceptable to include only English-language studies?

It is a limitation, and it should be reported as one. The Cochrane Handbook says searches should not be restricted by language or publication status, because such restrictions can introduce language bias or publication bias. If you restrict to languages your team can read, say so, report how many records were excluded for language at full-text stage, and mention it in your limitations. Do not apply the language filter in the database if you can avoid it, because then you cannot report the number.

Can I change my criteria after I start screening?

Clarifying an ambiguous criterion after piloting is normal. Changing criteria after seeing study results is a source of bias. In a systematic review, PRISMA 2020 (item 24c) asks authors to describe and explain any amendments to the information provided at registration or in the protocol. Keep a dated log of decisions either way.

Inclusion and Exclusion Criteria for a Literature Review: How to Write Them

Eligibility criteria decide which sources are allowed to shape your conclusions. Done well, they make a review reproducible: another person applying the same rules to the same records would select the same studies. Done badly, they are a list of vague preferences that cannot be applied consistently, or a filter chosen after the fact to keep the workload down.

One point of terminology first. In a primary study, inclusion and exclusion criteria define which participants can enrol. In a literature review they define which sources can be included. This guide is about the second kind.

Start from the question

Criteria are the review question restated as rules. If the question is loose, the criteria will be too. Question frameworks help because each element of the framework becomes a criterion.

Framework	Elements	Best suited to	Source
PICO (sometimes PICOS or PICOT)	Population, Intervention, Comparator, Outcome (+ Study design, Time)	Questions about the effect of an intervention or exposure	Cochrane Handbook, chapters 2 and 3
PCC	Population, Concept, Context	Scoping reviews	Munn et al. (2018), who recommend it for JBI scoping reviews
SPIDER	Sample, Phenomenon of Interest, Design, Evaluation, Research type	Qualitative and mixed-methods evidence	Cooke, Smith and Booth (2012)
PEO	Population, Exposure, Outcome	Questions about associations and risk factors where there is no intervention	A variant of PICO found in many library guides. It has no single founding paper that we could identify

The Cochrane Handbook notes that population, intervention and comparator usually translate directly into eligibility criteria. Outcomes are handled more cautiously. The Handbook says that "reporting of outcomes should rarely determine study eligibility", and that studies should not be excluded because they do not report results for an outcome they may have measured, in order to avoid bias from selective reporting. A thesis review that does require a relevant outcome should say so plainly.

If you have not settled the question yet, the research question guide comes first.

The categories most reviews need

Category	What to specify	Example wording
Population or sample	Age, condition, role, setting. Decide what to do with mixed samples	"Undergraduate students. Mixed samples included if results for undergraduates are reported separately or they make up at least 80% of the sample"
Intervention, exposure or phenomenon	A definition that distinguishes it from near neighbours	"Peer mentoring, defined as structured one-to-one or small-group support by a more senior student. Tutoring by paid staff excluded"
Comparator	If relevant, which comparisons count	"Any comparator, including no mentoring, waiting list or alternative support"
Outcomes	Which outcomes, measured how and when	"Retention to second year or course completion, from institutional records or self-report"
Study design or source type	Designs and document types you will accept	"Randomized and quasi-experimental studies, cohort studies. Editorials, opinion pieces and study protocols excluded"
Publication status	Peer-reviewed only, or also theses, reports, preprints, conference papers	"Peer-reviewed articles and doctoral dissertations. Conference abstracts excluded because they report insufficient detail to assess"
Date	Start and end, with a reason	"Published [year] onwards, because [the policy change, technology or earlier review that defines the period]"
Language	Languages the team can assess	"English, Spanish and Portuguese, the languages read by the authors"
Geography or context	Only if the question requires it	"Any country. Context recorded for subgroup comparison"

Every row should have a reason you could say out loud to an examiner. Excluding grey literature trades completeness for feasibility. The Cochrane Handbook warns that restricting by publication status creates a possibility of publication bias, which is why systematic reviews try to include unpublished work. If you exclude it, say why.

Write criteria that can be applied

A criterion works if two people reading the same abstract would make the same decision. Three tests help:

Is it observable in the report? "High-quality studies" cannot be judged at screening. "Studies with a control group" can.
Does it handle borderline cases? Mixed populations, multi-component interventions and studies reporting several outcomes are where screeners disagree. Decide in advance.
Is each exclusion criterion more than a mirror image? If the inclusion criterion says "adults", you do not need "children" as an exclusion criterion. Use exclusion criteria for things like a specific subgroup, a confounding co-intervention or a document type.

Worked example

This is an illustration of the format for an imagined review. The reasons in the last column are the kind you would write. They are not findings. Review question: does peer mentoring improve first-year retention among undergraduate students, compared with no mentoring?

	Inclusion	Exclusion	Reason
Population	First-year undergraduate students at degree-granting institutions	Samples limited to postgraduate or foundation-year students	The transition into degree study is the mechanism of interest
Intervention	Structured peer mentoring lasting at least four weeks	Programmes where mentors are paid staff. Mentoring delivered only as part of a bundled first-year seminar where its effect cannot be separated	Isolates the peer element
Comparator	No mentoring, waiting list or usual support	None	Any counterfactual is informative
Outcome	Retention, persistence or completion measured at or after the start of year two	Studies reporting only satisfaction or sense of belonging	The review is about staying enrolled
Design	Randomized, quasi-experimental and cohort designs with a comparison group	Single-group pre-post designs, qualitative studies, commentaries	A comparison group is needed to estimate an effect
Date	2010 to the search date	None	Earlier period covered by a previous review. Cite that review
Language	English or Spanish	None	Languages read by both screeners
Publication type	Journal articles, dissertations, institutional reports	Conference abstracts without a full paper	Insufficient detail for extraction

Notice that qualitative studies are excluded here although they would be central to a different question, such as how mentees experience mentoring. Criteria are correct or incorrect only in relation to a question.

Pilot before you screen everything

Take a random sample of 20 to 50 records from your search results and have every screener apply the criteria independently. Then compare.

Where you disagreed, was the criterion ambiguous or did someone misread the abstract? Reword ambiguous criteria.
Were there records nobody could classify? You may be missing a category.
If you calculate agreement, Cohen's kappa is a commonly used statistic. Pilot again after rewording if agreement was poor.

If you are working alone, pilot against yourself. Screen the sample, wait a few days, screen it again without looking, and compare. The records you changed your mind about show where the criteria are loose.

Apply them in a fixed order

At title and abstract stage, include when in doubt. The Cochrane Handbook says authors "should generally be over-inclusive at this stage". Abstracts often omit the detail you need, and a wrongly excluded record is gone for good, while a wrongly included one gets caught at full text.

At full-text stage, check criteria in the same order for every report and record the first one failed. That gives each excluded report a single reason, so the reasons listed in the PRISMA 2020 flow diagram add up to the number of reports excluded. A simple screening form is enough:

Report ID: ______   Screener: ____   Date: ________

1. Population: first-year undergraduates?            Yes / No / Unclear
2. Intervention: structured peer mentoring ≥ 4 wks?  Yes / No / Unclear
3. Design: has a comparison group?                   Yes / No / Unclear
4. Outcome: retention at or after start of year 2?   Yes / No / Unclear
5. Language: English or Spanish?                     Yes / No
6. Publication type eligible?                        Yes / No

Decision:  Include / Exclude / Discuss
If excluded, FIRST criterion failed (one only): ____
Notes (page numbers for the deciding information):

Put the cheapest checks first. Population and design can usually be determined from the methods section in a minute. Outcome timing may require reading the results.

Criteria for a narrative or thesis review

You do not need a registered protocol to benefit from explicit criteria. A short paragraph at the start of the chapter is enough. Example wording, with the parts you would replace in brackets:

This review covers peer-reviewed empirical studies and doctoral dissertations published between [year] and [year] on peer mentoring for first-year undergraduates. Studies of postgraduate mentoring and of staff-led tutoring were excluded. Searches were run in [databases] in [month and year] and were limited to [languages]. Seminal theoretical works published before [year] are discussed where they inform the framework.

The final sentence is an honest way to handle classic sources that fall outside the date range. Narrative reviews are allowed to do that as long as the exception is stated. The differences between review types are covered in systematic vs. scoping vs. narrative review.

Common mistakes

Writing criteria after screening, to describe what you happened to keep. Criteria are meant to be set before you see what they select.
Criteria that depend on results, such as including only studies that found a significant effect. This guarantees a biased answer.
Unjustified date limits. "Last ten years" needs a reason linked to the topic or to a previous review.
Filtering by language or document type inside the database and therefore being unable to report how many records that removed.
Using journal ranking as a quality criterion. Quartile or impact factor says something about the journal and little about an individual study. Appraise studies with a design-specific tool instead. Checking that a journal is genuine is a different matter, covered in how to evaluate journal quality.
Too many criteria. Each additional rule is another chance for screeners to disagree. If a characteristic does not change whether a study can answer the question, record it at extraction instead of screening on it.
No decision log. Borderline calls made in week two are forgotten by week ten. Keep a dated list.

When you search in Wonders, the year range, journal quartile and venue filters are set per search panel and stay attached to it. That makes it easier to report which limits you applied, and to rerun the same search without them to see what they removed.

References

Cooke, A., Smith, D., & Booth, A. (2012). Beyond PICO: the SPIDER tool for qualitative evidence synthesis. Qualitative Health Research, 22(10), 1435–1443. https://doi.org/10.1177/1049732312452938
Lefebvre, C., Glanville, J., Briscoe, S., Featherstone, R., Littlewood, A., Metzendorf, M.-I., Noel-Storr, A., Paynter, R., Rader, T., Thomas, J., & Wieland, L. S. Chapter 4: Searching for and selecting studies (last updated March 2025). In Cochrane Handbook for Systematic Reviews of Interventions, version 6.5.1. https://www.cochrane.org/authors/handbooks-and-manuals/handbook/current/chapter-04
McKenzie, J. E., Brennan, S. E., Ryan, R. E., Thomson, H. J., Johnston, R. V., & Thomas, J. Chapter 3: Defining the criteria for including studies and how they will be grouped for the synthesis (last updated August 2023). In Cochrane Handbook for Systematic Reviews of Interventions, version 6.5. https://www.cochrane.org/authors/handbooks-and-manuals/handbook/current/chapter-03
Methley, A. M., Campbell, S., Chew-Graham, C., McNally, R., & Cheraghi-Sohi, S. (2014). PICO, PICOS and SPIDER: a comparison study of specificity and sensitivity in three search tools for qualitative systematic reviews. BMC Health Services Research, 14, 579. https://doi.org/10.1186/s12913-014-0579-0
Page, M. J., McKenzie, J. E., Bossuyt, P. M., et al. (2021). The PRISMA 2020 statement. BMJ, 372, n71. https://doi.org/10.1136/bmj.n71
Munn, Z., Peters, M. D. J., Stern, C., Tufanaru, C., McArthur, A., & Aromataris, E. (2018). Systematic review or scoping review? Guidance for authors when choosing between a systematic or scoping review approach. BMC Medical Research Methodology, 18, 143. https://doi.org/10.1186/s12874-018-0611-x
Peters, M. D. J., Marnie, C., Tricco, A. C., Pollock, D., Munn, Z., Alexander, L., McInerney, P., Godfrey, C. M., & Khalil, H. (2020). Updated methodological guidance for the conduct of scoping reviews. JBI Evidence Synthesis, 18(10), 2119–2126. https://doi.org/10.11124/JBIES-20-00167
JBI. JBI Manual for Evidence Synthesis (see the chapter on scoping reviews). https://jbi-global-wiki.refined.site/space/MANUAL — listed as a pointer only: the site is a JavaScript application that serves no readable text to an automated request, so nothing on this page is sourced from it.