revisiaLog inSign up

Risk of bias figure generator

Traffic light plot and summary plot, robvis style. Free. Your data never leaves the browser.

What the risk of bias figure shows

A systematic review does not only say what each study found: it says how far that can be trusted. The risk-of-bias assessment is that judgement, domain by domain, and the figure is how it is shown at a glance. Cochrane requires it in its reviews and most journals expect it in any review with a meta-analysis.

There are two figures. The traffic light plot has one row per study and one column per domain, with a coloured dot at each crossing. The summary plot stacks, for each domain, the percentage of studies at each level. The first says where each study is weak; the second, where the body of evidence is.

Which instrument you need

The instrument follows the design of the included studies, not the review question. If you mix designs, make one figure per instrument.

  • RoB 2. Randomized controlled trials. Five domains and an overall judgement: low risk, some concerns, high risk.
  • RoB 1. The original Cochrane tool (2011) for trials, replaced by RoB 2. Seven domains, from random sequence generation to “other bias”; low, unclear or high risk.
  • ROBINS-I. Non-randomized studies of interventions. Seven domains; the levels are low, moderate, serious and critical.
  • ROBINS-I V2. Version 2 of ROBINS-I, still a draft (20 November 2025). Six domains: it drops deviations from intended interventions and moves classification ahead of selection.
  • ROBINS-E. Non-randomized studies of exposures. Seven domains; the levels are low, some concerns, high and very high.
  • QUADAS-2. Diagnostic accuracy studies. Four risk-of-bias domains and three applicability concerns, no overall judgement.
  • PROBAST. Studies that develop or validate prediction models. Four risk-of-bias domains (participants, predictors, outcome, analysis) and three applicability concerns; low, high or unclear.
  • Newcastle-Ottawa. Cohort and case-control studies. Three sections that award stars; the figure maps them to good, fair and poor quality.
  • Newcastle-Ottawa (cross-sectional). The adaptation by Herzog et al. for cross-sectional studies. Seven items and up to 10 stars; each cell says how many of the item’s stars the study got.
  • JBI. Checklist for analytical cross-sectional studies: yes, no, unclear or not applicable on each of its eight items.

With a page of their own: RoB 2 · ROBINS-I · QUADAS-2 · Newcastle-Ottawa

How to read the figure

Every dot is a judgement. Green with “+” is low risk; yellow with “–”, some concerns or moderate; red with “×”, high or serious risk; dark red with “!”, critical or very high; grey with “?”, no information. The symbol sits inside the dot so the figure survives black-and-white printing and colour blindness. If the journal asks, switch to a colour-blind safe palette (blue, yellow, vermilion) or to greyscale; the symbols stay the same.

Columns are named D1, D2… and the key under the figure says which domain each one is, in the instrument’s exact words. To read it without a key, choose “Full name of each domain” and each name runs up its own column. The last column, when the instrument defines one, is the study’s overall judgement.

In the summary plot every bar adds up to 100 % of the studies. With weighting on, each study counts according to its weight in the meta-analysis, and the bar says how much of the evidence is at risk rather than how many papers.

How to fill it in

  1. Choose the instrument and, if the journal asks, the language of the figure.
  2. Add one row per study or paste the table from your spreadsheet. The downloadable template has robvis’s columns: Study, D1…Dn, Overall and Weight.
  3. Mark the level of each domain. The cell takes the colour of the judgement and the figure redraws at once.
  4. Fill in the overall judgement by hand or with the button that applies the instrument’s rule.
  5. Download the traffic light plot, the summary plot or both.

Exporting for a journal

Use PDF or SVG when the journal asks for vector artwork; both keep the text sharp at any size. When the submission system wants an image, download PNG or TIFF at the journal’s column width (8.5, 12 or 17.5 cm) and 300 or 600 dpi: the resolution is written into the file, so the journal reads the right size without anyone resizing it. The tool warns you if the text would print below 6 pt. The CSV keeps the table with robvis’s columns, to load it here again or open it in R; without an account, the heading of its first column carries the revisia line, which robvis does not read.

What you download carries a revisia mark at the bottom right, outside the figure. Sign in to your revisia account with a verified email and the figure comes out clean, in the preview and in every format. The account is free.

Frequently asked questions

Which instrument should I use for which study design?

RoB 2 for randomized controlled trials (RoB 1 if your review or journal still asks for the 2011 tool). ROBINS-I for non-randomized studies of interventions (treated cohorts, before-after designs); its version 2, still a draft, is here too. ROBINS-E for non-randomized studies of exposures (environmental, occupational, dietary). QUADAS-2 for diagnostic accuracy studies. PROBAST for prediction model studies. Newcastle-Ottawa for cohort and case-control studies; for cross-sectional studies, its adaptation by Herzog et al. or the JBI checklist. If your instrument is not listed, the custom instrument lets you name 2 to 12 domains.

What do the colours and symbols mean?

Green with “+” is low risk (or “yes” in a checklist). Yellow with “–” is some concerns, moderate or unclear. Red with “×” is high or serious risk. Dark red with “!” is critical or very high, a level only ROBINS-I and ROBINS-E have. Grey with “?” is no information. The symbol sits inside the dot so the figure still reads in black and white and for colour-blind readers. These are the classic colours; you can also pick a colour-blind safe palette or greyscale, and the symbols stay the same.

Is this the same figure robvis makes?

It is the same kind of figure: the traffic light plot and the summary plot made standard by the robvis package of McGuinness and Higgins. The arrangement, the levels and the data template follow robvis so you can move your data either way. The colours are revisia’s, and every dot carries a symbol, which robvis offers too.

How do I cite the figure and the instruments?

Cite the instrument you used: RoB 2 is Sterne JAC et al. BMJ 2019;366:l4898; RoB 1 is Higgins JPT et al. BMJ 2011;343:d5928; ROBINS-I is Sterne JAC et al. BMJ 2016;355:i4919; ROBINS-E is Higgins JPT et al. Environ Int 2024;186:108602; QUADAS-2 is Whiting PF et al. Ann Intern Med 2011;155:529-36; PROBAST is Wolff RF et al. Ann Intern Med 2019;170:51-8; the Newcastle-Ottawa Scale is Wells GA et al. (Ottawa Hospital Research Institute), and its adaptation for cross-sectional studies Herzog R et al. BMC Public Health 2013;13:154. To acknowledge the figure type, cite robvis: McGuinness LA, Higgins JPT. Res Synth Methods 2021;12:55-61. This tool itself needs no citation.

Does my data leave the browser?

No. The table, the figure and every file are generated on your device. The draft is kept in your browser’s local storage, and the share link carries the table inside the address itself, after the hash, which browsers never send to any server.

Why is there a watermark and how does it go away?

What you download carries a revisia mark at the bottom right, outside the figure; the CSV carries it in the heading of the first column, which robvis does not read. Sign in to your revisia account with a verified email and the figure comes out clean, in the preview and in every format. The account is free and there is no paid version of this tool.

What about the overall judgement in QUADAS-2 and JBI?

QUADAS-2 defines no overall judgement: it asks for each domain to be reported separately, and the figure does that, with the three applicability questions in a block of their own. The JBI checklist ends in a decision (include, exclude, seek further information), not a risk level, so it has no overall column either. In RoB 2, ROBINS-I and ROBINS-E the “Fill in the overall judgement” button applies the instrument’s rule, the worst domain decides, and you can then change it. Bear in mind that RoB 2 and ROBINS-E allow several domains at “some concerns” to justify an overall judgement of high: that call is yours, the button does not make it.

How do Newcastle-Ottawa stars come in?

The figure uses the three AHRQ quality bands (good, fair, poor) because a star count has no colour of its own. Pick the band in each cell, or paste the star count of each section: selection 3–4 good, 2 fair, 0–1 poor; comparability 1–2 good, 0 poor; outcome or exposure 2–3 good, 0–1 poor. Only selection can be “fair”: under the AHRQ standard a study is good with 3–4 stars in selection, 1–2 in comparability and 2–3 in outcome; fair with 2 in selection and the same in the other two; and poor with 0–1 in selection, 0 in comparability or 0–1 in outcome. The conversion happens as you paste.

The figure is the last step of the assessment. revisia does the ones before it.

revisia assesses the risk of bias of every included study, domain by domain, and points to the sentence in the paper that supports each judgement so you can confirm or change it. The figure draws itself from what you decide. The account is free; the AI assessment is part of the paid plans.

Create a free account

Domains and judgement levels are those published by each instrument’s authors: RoB 2, Sterne JAC et al. BMJ 2019;366:l4898; RoB 1, Higgins JPT et al. BMJ 2011;343:d5928; ROBINS-I, Sterne JAC et al. BMJ 2016;355:i4919; ROBINS-I V2, draft of 20 November 2025 (riskofbias.info); ROBINS-E, Higgins JPT et al. Environ Int 2024;186:108602; QUADAS-2, Whiting PF et al. Ann Intern Med 2011;155:529-36; PROBAST, Wolff RF et al. Ann Intern Med 2019;170:51-8; Newcastle-Ottawa Scale, Wells GA et al., Ottawa Hospital Research Institute; its adaptation for cross-sectional studies, Herzog R et al. BMC Public Health 2013;13:154; JBI checklist for analytical cross-sectional studies, JBI. The traffic light and summary plots follow the format of the robvis package: McGuinness LA, Higgins JPT. Risk-of-bias VISualization (robvis): an R package and Shiny web app for visualizing risk-of-bias assessments. Res Synth Methods 2021;12:55-61. This tool is independent: it is not affiliated with, endorsed by or maintained by Cochrane, the instruments’ authors or robvis. Your data never leaves your browser. Other free tools: all tools · PRISMA flow diagram.