Published
The SANRA scale is a six-item tool that scores the quality of a narrative review from 0 to 12. Editors use it to triage submissions, reviewers use it to structure a report, and authors can use it to audit a draft before anyone else reads it. This guide explains the six items, then applies every one to a real open-access narrative review, names the section where each answer sits, and marks the rows where two careful raters could disagree.
The scale was published by Baethge, Goldbeck-Wood and Mertens in Research Integrity and Peer Review in 2019, and its form and instructions are free to reuse under a Creative Commons licence. If you are still choosing between review types, our guide to narrative versus systematic literature searches covers the search side first, and our scoping, systematic and narrative review comparison explains how a narrative review sits beside a scoping or systematic review. This page covers the scoring itself.
In Simple Words
The SANRA scale asks six questions about a narrative review. Does it show why readers need it, does it state a clear aim, does it say how the literature was found, are the key statements referenced, is the reasoning built on suitable evidence, and are the data presented properly. Each question scores 0, 1 or 2, so the total runs from 0 to 12. Below, the six questions are applied to a real review so you can see what each score looks like on the page.
What the SANRA scale measures and who built it
Three editors at Deutsches Ärzteblatt, the journal of the German Medical Association, developed the SANRA scale between 2010 and 2017. Two earlier seven-item versions were tested first, one on 10 narrative reviews in neurology and one on 12 consecutive submissions to the journal. The current version was revised in 2014, and the authors dropped an item on writing and accessibility because raters differed too much on it. What remains is a scale that covers the importance of the article, its aims, the literature search, referencing, scientific reasoning and the presentation of data.
The published field test used the first 30 consecutive review manuscripts without a systematic literature search that reached the journal in 2015. Three editors with at least 10 years of experience each scored every manuscript independently, which gave 90 ratings. The mean sum score was 6.0 out of 12 (SD 2.6, range 1 to 12). Internal consistency measured by Cronbach alpha was 0.68, and inter-rater reliability was an intra-class correlation of 0.77 (95% CI 0.57 to 0.88). Raters disagreed most often on items 1 and 4. In a post hoc analysis, accepted manuscripts averaged 6.6 (n = 17) and rejected ones 5.1 (n = 13), a point biserial correlation of 0.37.
The authors are open about the limits. There are no established cut-offs that travel across journals, although in their experience a score of 4 or below indicates very poor quality. SANRA is a critical appraisal tool, not a reporting guideline, and it cannot judge originality, plausibility or the correctness of the content. That scope matters for how you read any SANRA score, including the one below.
The six SANRA scale items and what each score means
Figure 1 shows the scale as published. Each item has three anchors, and the instructions document asks raters to use the extreme scores of 0 and 2 freely rather than saving them for perfect or hopeless articles. Ratings should cover the whole manuscript, including the abstract.
| Item | What it asks | What a score of 2 looks like | What a score of 0 looks like |
|---|---|---|---|
| 1 | Justification of the importance of the article for its readers | The clinical problem and the unanswered question or evidence gap are outlined thoroughly. | The importance is not justified. |
| 2 | Statement of concrete aims or questions | One or more specific aims or questions are stated clearly. | No aims or questions are formulated. |
| 3 | Description of the literature search | Search terms and the types of literature included are specified. A full systematic search with flowcharts is not required. | The search strategy is not presented. A brief mention scores 1. |
| 4 | Referencing | Statements the argument depends on are backed by references in all or almost all cases. Statements with uncontroversial face validity can be exempt. | Key statements are not supported by references. Inconsistent support scores 1. |
| 5 | Scientific reasoning | Key arguments rest on appropriate evidence, with the study design named and levels of evidence given where they exist. | The article’s point is not based on appropriate arguments. Selective use of evidence scores 1. |
| 6 | Appropriate presentation of data | Outcome data central to the argument are selected and presented correctly, for example absolute as well as relative risk, and effect sizes with confidence intervals. | Data are presented inadequately. Often not in the most appropriate way scores 1. |
Items 5 and 6 are easy to confuse. Item 5 is about the use of evidence and of types of evidence in the argument. Item 6 is about the selection and presentation of the concrete outcome numbers. A review can reason well from the right kinds of study and still leave its readers without the figures they need, and the example below shows exactly that split.
How we chose and checked the example
We searched PubMed for narrative reviews published in 2026 that offer free full text and mention SANRA in their methods, and we kept one whose licence we could confirm three ways. Crossref lists the version of record under CC BY 4.0, Europe PMC flags the article as CC BY, and the licence statement on the publisher page names the Creative Commons Attribution License.
Each row of the scoring table below carries the section where the answer sits, the evidence on the page, the score, and a call. Clear means the score follows directly from the anchors. Judgement call means two careful readers could score the row differently. Every quotation is 25 words or fewer, sits in quotation marks, and was checked against the publisher’s full text. We scored the Corrected Version of Record (23 October 2026; correction notice doi:10.3389/fped.2026.1994417) as read on 1 November 2026. Frontiers lists a correction to this article dated 23 October 2026, and the Europe PMC full-text copy we first retrieved differed from the corrected page in its reference list and in how it described the cost-effectiveness evidence. A SANRA score belongs to a specific version, so record the date and version whenever you score a paper.
The scoring was done by Editverse from the article alone. The authors were not consulted, and the scores are a reading of the text against the published anchors, not a verdict on the clinical conclusions.
The review behind this example
The article is “The case for broader use of HOMA-IR in pediatric metabolic assessment: a narrative review” by Snouda, published in Frontiers in Pediatrics on 28 August 2026 (PMID 42729805). It argues that the homeostatic model assessment for insulin resistance, HOMA-IR, deserves wider use in paediatric metabolic assessment, and it has 9 numbered sections, 3 tables, 1 figure and 35 references. Its abstract states that the review was conducted in accordance with the SANRA framework.
That makes it a fair teaching case. A review that names SANRA invites the question of how it scores, and the answer shows what the scale rewards and where even a careful author leaves points on the table. It also has a real methods paragraph, three tables, a conclusion that lists its own limits, and a reference list that can be checked, which are the features SANRA needs in order to be applied at all.
SANRA scale scoring of the example, item by item
Table 2 gives the full scoring. Page pointers use the section numbers of the published article.
| Item | Question | Where in the article | Evidence | Score | Call |
|---|---|---|---|---|---|
| 1 | Importance for readers | Abstract, sections 1.2 and 1.2.1 | The abstract opens with the problem, that body mass index and fasting glucose “may systematically fail to detect the earliest phase of metabolic dysfunction”. Section 1.2 explains why each marker falls short, and section 1.2.1 names the gap, that “no prospective population-level screening trial has demonstrated that routine HOMA-IR screening improves clinical outcomes”. | 2 | Clear |
| 2 | Concrete aims or questions | Abstract, section 1.1, section 1.3 | The abstract states the aim “to evaluate the evidence for the Homeostatic Model Assessment for Insulin Resistance (HOMA-IR) as a potential first-line tool in pediatric metabolic assessment”. Section 1.1 lists six thematic domains, and section 1.3 poses the central question of universal versus targeted screening. | 2 | Clear |
| 3 | Description of the literature search | Section 1.1 | The methods name three sources, “A systematic literature search was conducted in PubMed/MEDLINE, Scopus, and Google Scholar”, list nine search phrases, give four inclusion criteria and three exclusions, and add hand-searching of reference lists and four guideline bodies. Weighting was “toward studies published from 2010 onwards”. The search date and record counts are not given, which SANRA does not require. | 2 | Clear |
| 4 | Referencing of key statements | Throughout, with gaps in sections 5.2.5, 5.3 and 7.2 | Most key numbers carry a citation, for example the 20% to 40% assay difference (reference 26), the lifetime cost of type 2 diabetes (reference 21) and the prepubertal cut-off of 2.5 (references 15 and 16). The drug-approval statements in section 7.2, such as “Metformin is the only pharmacological agent currently approved for use in children aged ≥10 years with established T2D in most jurisdictions”, carry no reference. | 2 | Judgement call |
| 5 | Scientific reasoning | Sections 1.1, 3, 5.4, 6.3, 7.1 and the Table 3 caption | The authors label evidence types as they go. The methods say “Where the evidence is primarily mechanistic or derived from modelling studies, this is explicitly noted”. Section 6.3 describes a “small pilot randomised trial” and adds that “these findings require confirmation in larger trials”. Table 3 is captioned as an expert-opinion framework. Table 1 rates the paediatric evidence as “Strong”, citing references 15 to 18, but the text does not describe the design or size of each study, and the abstract of reference 15 reports 208 obese children. | 2 | Judgement call |
| 6 | Presentation of data | Section 4.2, Tables 1 and 2, sections 5.1, 5.2.3 and 6.1 | Some figures are given, including a 20% to 40% assay difference, a 25% to 50% fall in insulin sensitivity across puberty and a lifetime cost above $85,000. The central screening claim, however, is stated as “Consensus from multiple large-scale cohorts suggests that values >2.5 consistently correlate with increased metabolic risk in prepubertal children”, with no sensitivity, specificity or area under the curve. The abstract of reference 15 reports cut-offs from 2.22 to 5.22 by sex and pubertal stage, with sensitivities of 56% to 100% and specificities of 42% to 93%. The authors also state that no completed trial of HOMA-IR-guided screening has shown benefit, which limits what could be shown. | 1 | Judgement call |
| Total | Sum of items 1 to 6 | Defensible range across stricter and more lenient readings is 9 to 12. | 11 of 12 | Three judgement calls |
What a total of 11 out of 12 tells you
A total of 11 sits far above the field-test mean of 6.0, and it fits what the article does well. It states why the topic matters, says what it will do, describes its search in enough detail to repeat the search terms, and keeps telling the reader which claims rest on weak evidence. Figure 2 sets the item scores beside the field-test means.
The one item where the example loses a point, item 6, is also the item with the lowest mean in the field test (0.81), followed by item 5 (0.83) and item 3 (0.84). The authors of the scale found the same pattern in routine submissions, so a weak data presentation is not a quirk of this paper. It is the part of narrative reviewing where authors most often stop short of the numbers that readers need.
How the total changes under a stricter reading
The three judgement calls move the total in both directions, which is why the table lists a range rather than a single answer.
- Item 4 could be 1. A rater who counts the unreferenced drug-approval statements in section 7.2 as key to the argument would score referencing as inconsistent. A rater who treats them as peripheral to a review about screening would keep the 2.
- Item 5 could be 1. A rater who wants the design and size of each cited study named in the text, not only in the reference list, would score the reasoning as less than thorough. A rater who values the repeated, explicit labelling of mechanistic, modelling and expert-opinion evidence would keep the 2.
- Item 6 could be 2. A rater who accepts that the authors state no outcome trial exists, and who judges the numbers given as sufficient for a review of this scope, would score data presentation as appropriate.
Strict on items 4, 5 and 6 gives 9, and lenient on item 6 gives 12. That spread is normal for the scale. The field test reported an intra-class correlation of 0.77 with a lower confidence limit of 0.57, and the authors recommend that raters practise on a few manuscripts and use the instructions document before scoring in earnest. When two raters differ, discuss the anchors for the disputed item rather than averaging blindly.
Six habits that raise a SANRA scale score before submission
Authors can run the scale on their own draft and fix the lowest items before submission. The authors of the scale encourage this use, and Gasparyan and colleagues and Pautasso offer complementary writing advice for narrative reviews.
- Justify the topic in two sentences with a reference. State the clinical problem and the unanswered question, as the example does in section 1.2.1 (item 1).
- Write the aim as a question or a list of topics. One sentence in the abstract and one in the introduction is enough (item 2).
- Put the search in a short methods paragraph. Give the sources, the exact terms, the date, the inclusion and exclusion rules and the types of literature, which is all item 3 asks for (item 3).
- Audit your key statements. Highlight every sentence the argument depends on and check that each has a reference, including regulatory and drug-approval statements (item 4).
- Name the evidence type at the point of the claim. Say randomised trial, cohort, animal model, modelling study or expert opinion, with the size where it matters (item 5).
- Give the outcome numbers that readers will ask for. Report absolute as well as relative differences and put an interval on every effect size. For a test, give sensitivity and specificity with the cut-off they refer to (item 6).
Our guide to writing a literature review walks through the same discipline for a thesis chapter.
SANRA scale versus AMSTAR 2, CASP and PRISMA
SANRA fills a gap that the better-known instruments leave open. The scale paper contrasts its six items with the 16 items of AMSTAR 2 for systematic reviews and the 11 items of the CASP randomised trial checklist. Pick the instrument by the document in front of you.
| Tool | Designed for | Role | Size |
|---|---|---|---|
| SANRA scale | Narrative (non-systematic) reviews | Critical appraisal by editors, reviewers, readers and authors | 6 items scored 0 to 2 |
| AMSTAR 2 | Systematic reviews of healthcare interventions | Critical appraisal with confidence ratings | 16 items |
| CASP checklists | Single studies, by design | Critical appraisal of one study at a time | 11 items for the randomised trial checklist |
| PRISMA 2020 | Systematic reviews and meta-analyses | Reporting guideline, what to write | 27 items |
| PRISMA-ScR | Scoping reviews | Reporting guideline, what to write | 22 items (20 essential, 2 optional) |
Appraisal tools judge a finished document, while reporting guidelines tell authors what to include. SANRA is the first kind, so it works best alongside a good writing plan rather than instead of one.
Frequently asked questions about the SANRA scale
What does SANRA stand for?
SANRA stands for Scale for the Assessment of Narrative Review Articles. It has six items, each scored 0, 1 or 2, so the maximum is 12.
What is a good SANRA score?
No validated cut-off exists across journals. In the field test the mean was 6.0 out of 12, and the authors say a score of 4 or below indicates very poor quality in their experience. Compare manuscripts scored by the same raters in the same setting rather than across settings.
Is the SANRA scale a reporting guideline?
No. The authors describe it as a critical appraisal tool and state that it is not a reporting guideline. It does not cover originality, tables and figures, or whether the content is correct.
Does a narrative review need a systematic search to score 2 on item 3?
No. The instructions say a score of 2 does not need the detail of a systematic review search, but it does need the search terms and the types of literature included. A manuscript that mentions its search only briefly scores 1, and one that does not mention its methods scores 0.
Can authors score their own manuscript with SANRA?
Yes. The authors of the scale encourage authors to use it while preparing narrative reviews. Scoring a draft against the anchors, ideally with a colleague, shows which of the six items needs work before submission.
Where can readers get the SANRA form and instructions?
The form and the explanations and instructions document appear as figures in the original article, and the authors host the form on the Deutsches Ärzteblatt website. The article is published under CC BY 4.0.
Free Research Tools from Editverse
- GraphPro: build publication-ready charts, 25 types, with built-in t-test, ANOVA, correlation, and chi-square p-values. Free, no signup.
- StatPro: free statistical calculators and a test-selector tool covering group comparisons, categorical outcomes, time-to-event analysis, and diagnostic accuracy.
- Flowchart Pro: generate PRISMA 2020, PRISMA-ScR, CONSORT, STROBE, STARD, MOOSE, and ARRIVE flow diagrams directly in your browser.
Professional support for a narrative review that scores well
Two of the three lowest-scoring items in the field test, the search description and the presentation of data, are also the two that benefit most from a second pair of hands. Two Editverse services map onto them.
Narrative review
$1,000
Narrative review writing with topic scoping, a non-systematic literature search, critical analysis, thematic organisation, gap identification and journal recommendations. 4 to 6 weeks.
Search documentation
$400
A database search by information specialists across five major databases, with search strategies, search date and yield documentation, and a deduplicated reference library. 1 to 2 weeks.
Our work aligns with COPE, ICMJE and GPP 2022, and follows institutional and country-specific ethics requirements.
Using the SANRA scale well
The SANRA scale rewards a narrative review that explains why it exists, says what it will do, shows how the literature was found, references what matters, labels its evidence and gives readers the numbers. The example review earns 11 of 12 on a careful reading, loses its one point on data presentation, and shows that three of the six rows are judgement calls even for a strong paper. Score your own draft early, fix the lowest item first, and record the version you scored. For the sibling tools that appraise other designs, return to the AMSTAR 2 guide and the CASP guide.
References
- Baethge C, Goldbeck-Wood S, Mertens S. SANRA, a scale for the quality assessment of narrative review articles. Res Integr Peer Rev. 2019;4:5. doi:10.1186/s41073-019-0064-8. PMID 30962953. PMCID PMC6434870.
- Snouda S. The case for broader use of HOMA-IR in pediatric metabolic assessment: a narrative review. Front Pediatr. 2026;14:1883916. doi:10.3389/fped.2026.1883916. PMID 42729805. PMCID PMC13563044.
- Kurtoglu S, Hatipoglu N, Mazicioglu M, Kendirci M, Keskin M, Kondolot M. Insulin resistance in obese children and adolescents. HOMA-IR cut-off levels in the prepubertal and pubertal periods. J Clin Res Pediatr Endocrinol. 2010;2(3):100-106. doi:10.4274/jcrpe.v2i3.100. PMID 21274322. PMCID PMC3005684.
- Gasparyan AY, Ayvazyan L, Blackmore H, Kitas GD. Writing a narrative biomedical review. Considerations for authors, peer reviewers, and editors. Rheumatol Int. 2011;31(11):1409-1417. doi:10.1007/s00296-011-1999-3. PMID 21800117.
- Pautasso M. Ten simple rules for writing a literature review. PLoS Comput Biol. 2013;9(7):e1003149. doi:10.1371/journal.pcbi.1003149. PMID 23874189. PMCID PMC3715443.