Assess risk of bias in prognostic factor studies using the QUIPS framework (Hayden et al., 2013) with 6 domains, traffic-light visualization, summary bar charts, and publication-ready PNG export.
Add your prognostic factor studies and enter their names. Click each colored circle to cycle through judgments: + Low risk, ! Moderate risk, − High risk, ? N/A. Each domain should be assessed using the QUIPS prompting items. When done, export the table as a high-resolution PNG for your manuscript.
Load sample data to see how the tool works, or clear all fields to start fresh.
| Study | D1 Study Participation | D2 Study Attrition | D3 Prognostic Factor Measurement | D4 Outcome Measurement | D5 Study Confounding | D6 Statistical Analysis and Reporting | |
|---|---|---|---|---|---|---|---|
QUIPS (Hayden JA et al., 2013) • Generated with Research Gold
Click Add Study to create a row for each prognostic factor study included in your systematic review. Enter the study identifier (e.g., Author Year) and your progress will be automatically saved to browser local storage so you can return later without losing work.
Evaluate whether the study population was adequately described, whether the sample was representative of the population of interest, and whether inclusion and exclusion criteria were appropriate. Assign Low, Moderate, or High risk of bias based on the prompting items for this domain.
Assess the completeness of follow-up data. Consider the proportion of participants with complete data, whether reasons for loss to follow-up were reported, and whether baseline characteristics differed between completers and non-completers. High attrition (typically above 20%) without adequate handling raises concern.
For the prognostic factor domain, evaluate whether the factor was measured validly, reliably, and consistently across participants. For outcome measurement, assess whether the outcome was defined clearly and measured using a valid method with appropriate blinding to the prognostic factor status.
For confounding, evaluate whether important confounders were identified, measured, and accounted for in the analysis. For statistical analysis, assess whether the analytical strategy was appropriate, whether model assumptions were satisfied, and whether results were reported completely including effect estimates and precision.
Review the traffic light table showing per-study, per-domain judgments and the summary bar chart displaying the proportion of studies at each risk level across all six domains. Download both visualizations as high-resolution PNGs suitable for journal submission.
Need this done professionally? Get a complete systematic review or meta-analysis handled end-to-end.
Get a Free QuoteQUIPS evaluates Study Participation, Study Attrition, Prognostic Factor Measurement, Outcome Measurement, Study Confounding, and Statistical Analysis and Reporting. Each domain has its own set of prompting items that guide assessors through the evaluation. Unlike numeric scales that collapse quality into a single score, QUIPS preserves domain-level information that is more useful for sensitivity analyses and GRADE assessments.
QUIPS is designed for studies investigating whether a factor predicts an outcome over time, not whether an exposure causes an outcome. If the study question is causal (does smoking cause lung cancer?), ROBINS-I or RoB 2 may be more appropriate. If the question is prognostic (does tumor size predict survival?), QUIPS provides the relevant domain structure. The distinction is subtle but affects which biases are most important to assess.
Prognostic studies often involve long follow-up periods, making loss to follow-up a pervasive threat. When attrition exceeds 20% and is differential (related to the prognostic factor or outcome), the remaining sample may no longer represent the original cohort. QUIPS prompting items for the Attrition domain ask whether attrition reasons were reported, whether completers differed from non-completers, and whether missing data were handled appropriately.
Unlike randomized trials where randomization balances confounders, observational prognostic studies must identify, measure, and statistically adjust for all important confounders. The QUIPS Confounding domain asks whether key confounders were defined, measured validly, and accounted for in the analysis. Residual confounding from unmeasured variables remains a limitation that cannot be fully addressed through statistical adjustment alone.
The GRADE framework adapted for prognosis (Iorio et al., 2015) starts with high certainty for prospective cohort studies and allows downgrading based on risk of bias, inconsistency, indirectness, imprecision, and publication bias. QUIPS domain ratings directly inform the risk of bias downgrading decision. A pattern of High risk across the Confounding and Attrition domains may justify downgrading certainty by one or two levels.
Use QUIPS when included studies examine the independent association between a prognostic factor and an outcome (e.g., does baseline depression severity predict treatment response?). Use RoB 2 for randomized trials of interventions. Use ROBINS-I for non-randomized studies that compare the effects of two or more interventions. The study design matters less than the research question: a cohort study asking a prognostic question should be assessed with QUIPS regardless of whether it also reports intervention effects.
Systematic reviews of prognostic factors require specialized tools that address the unique methodological challenges of prognosis research. The QUIPS tool (Quality In Prognosis Studies) provides a structured, domain-based framework developed by Hayden et al. (2013) and endorsed by the Cochrane Prognosis Methods Group. Its six domains (Study Participation, Attrition, Prognostic Factor Measurement, Outcome Measurement, Confounding, and Statistical Analysis) each target a distinct source of bias, with prompting items that help two independent reviewers reach reproducible judgments. The tool was validated across multiple prognostic systematic reviews (Huguet et al., 2013), demonstrating acceptable inter-rater reliability when assessors receive adequate training.
The Study Participation domain evaluates whether the enrolled sample adequately represents the population of interest. Selection bias occurs when the study sample differs systematically from the target population in ways that affect the prognostic association. The Study Attrition domain is particularly critical in prognosis research because longitudinal studies often span years or decades, creating multiple opportunities for participant dropout. When attrition is related to both the prognostic factor and the outcome (informative censoring), the observed association may be biased in either direction. Sensitivity analyses using multiple imputation or inverse probability weighting can partially address this concern.
Prognostic factor measurement must be valid, reliable, and consistently applied across all participants at a relevant time point. If the factor is measured with substantial error (low reliability), the observed association will be attenuated toward the null, potentially causing reviewers to underestimate the true prognostic value. The outcome measurement domain similarly requires that the outcome is defined clearly, measured with a validated instrument, and assessed without knowledge of the prognostic factor status. Differential misclassification of the outcome based on knowledge of the prognostic factor can bias the association in either direction.
The confounding domain represents the most complex assessment in QUIPS. Unlike randomized trials where randomization balances measured and unmeasured confounders, observational prognostic studies must identify all important confounders a priori and adjust for them analytically. Key confounders typically include age, sex, disease severity, comorbidities, and treatment received. When studies fail to adjust for known confounders, the observed prognostic association may reflect confounding rather than a true independent predictive relationship. Directed acyclic graphs (DAGs) can help review teams identify which variables are true confounders versus mediators or colliders.
QUIPS assessments feed directly into the GRADE framework adapted for prognosis (Iorio et al., 2015; Huguet et al., 2013), where risk of bias is one of five domains determining overall certainty of evidence. Reviewers should perform sensitivity analyses excluding studies with high risk on key domains, particularly confounding and attrition, to test the robustness of their pooled estimate. The forest plot generator can visualize these subgroup analyses, showing how the meta-analytic estimate changes when high-risk studies are removed. When heterogeneity is substantial, explore whether methodological quality differences explain variation across studies using the leave-one-out sensitivity analysis tool.
Selecting the correct bias assessment tool is essential for methodological rigor. For randomized trials, use RoB 2. For non-randomized intervention studies, use ROBINS-I. For cohort and case-control study quality scoring, the Newcastle-Ottawa Scale offers a star-based system. QUIPS fills the specific niche of prognostic factor research, providing domain-level assessments that are more informative than a single numeric score and more aligned with the causal questions inherent to prognosis. When your systematic review includes both prognostic and etiologic questions, consider using QUIPS for the prognostic analyses and ROBINS-I for the causal inference analyses, documenting the rationale for tool selection in your methods section.
QUIPS (Quality In Prognosis Studies) is a validated tool developed by Hayden et al. (2013) for assessing risk of bias in studies that investigate prognostic factors. Use QUIPS when your systematic review includes cohort studies, case-control studies, or other observational designs that examine the association between a prognostic factor (exposure) and a health outcome over time. It is specifically designed for prognosis research rather than intervention studies.
QUIPS evaluates bias across six domains: (1) Study Participation, which assesses selection and enrollment methods; (2) Study Attrition, which evaluates loss to follow-up and completeness of data; (3) Prognostic Factor Measurement, which judges whether the factor of interest was measured validly and reliably; (4) Outcome Measurement, which assesses whether the outcome was defined and measured appropriately; (5) Study Confounding, which evaluates whether important confounders were identified and accounted for; and (6) Statistical Analysis and Reporting, which judges whether the analytical methods were appropriate and fully reported.
QUIPS is purpose-built for prognostic factor studies that examine associations between patient characteristics and outcomes over time. ROBINS-I assesses risk of bias in non-randomized studies of interventions (treatments), while RoB 2 evaluates randomized controlled trials. The key distinction is the study question: if the study asks whether a factor predicts an outcome (prognosis), use QUIPS. If it asks whether an intervention causes an effect, use ROBINS-I or RoB 2 depending on the design.
QUIPS does not prescribe a formal algorithm for an overall judgment in the same way ROBINS-I does. However, most systematic reviews that use QUIPS assign an overall rating based on the pattern of domain judgments. A common approach is to rate a study as high risk overall if two or more domains are rated high, moderate if one domain is high or multiple domains are moderate, and low if most or all domains are rated low. You should document your decision rule in your review protocol.
Each QUIPS domain includes a set of prompting items (also called consideration items) that guide the assessor through the evaluation. These items highlight specific methodological aspects to consider within each domain. For example, in the Study Participation domain, prompting items address the adequacy of the study population description, the representativeness of the sample, the inclusion and exclusion criteria, and the baseline characteristics. Assessors use these items to inform their overall domain judgment of Low, Moderate, or High risk of bias.
The GRADE framework for prognosis (Iorio et al., 2015) uses QUIPS assessments as one of the factors when rating the overall certainty of prognostic evidence. Within GRADE for prognosis, risk of bias assessed by QUIPS can lead to downgrading the certainty of evidence from high to moderate, low, or very low. Systematic reviews of prognostic factors should present both the individual study QUIPS assessments and the overall GRADE rating to give readers a complete picture of evidence quality.
Assessing randomized trials instead? Use our Cochrane RoB 2 assessment tool with 5 domains and traffic-light visualization. For non-randomized intervention studies, apply the ROBINS-I bias assessment tool with 7 domains. For cohort and case-control study quality scoring, use the Newcastle-Ottawa Scale calculator. When you are ready to visualize your pooled results, generate publication-ready figures with our forest plot generator.
Reviewed by
Dr. Sarah Mitchell holds a PhD in Biostatistics from Johns Hopkins Bloomberg School of Public Health and has over 15 years of experience in systematic review methodology and meta-analysis. She has authored or co-authored 40+ peer-reviewed publications in journals including the Journal of Clinical Epidemiology, BMC Medical Research Methodology, and Research Synthesis Methods. A former Cochrane Review Group statistician and current editorial board member of Systematic Reviews, Dr. Mitchell has supervised 200+ evidence synthesis projects across clinical medicine, public health, and social sciences. She reviews all Research Gold tools to ensure statistical accuracy and compliance with Cochrane Handbook and PRISMA 2020 standards.
We conduct full risk of bias assessments, GRADE evaluations, and complete systematic reviews with rigorous methodology that satisfies peer reviewers. Average turnaround: 2-4 weeks.