Why English? | Reading QUADAS-C Comparative Diagnostic Accuracy Through Paired Tests, Allocation and Fair Comparisons

People searching for QUADAS-C, comparative diagnostic accuracy or how to compare two medical tests need precise language for a fair comparison. English matters because a fully paired design is not the same as two unrelated study groups, allocation concealment is not the same as blinding, and a difference in accuracy can reflect unequal conduct rather than a genuinely better test.

The official University of Bristol QUADAS-C page, checked on 5 November 2026, says QUADAS-C assesses risk of bias in primary studies comparing two or more index tests. It is not intended for indirect comparisons between separate single-test studies and does not assess applicability. The page also states that QUADAS-C was developed with QUADAS-2, cannot be used alone and now should be adapted alongside QUADAS-3 using the current explanation-and-elaboration guidance.

The official QUADAS-C tool distinguishes fully paired, randomised, partially paired and unpaired non-randomised designs, then examines patient selection, index-test conduct, the reference standard, flow and timing. This article owns the comparative risk-of-bias intent. The existing QUADAS-3 article owns the current main diagnostic-accuracy appraisal; STARD owns reporting completeness. None of these tools proves patient benefit, affordability or individual suitability.

• Define the exact tests, patients, target condition and comparison.
• Identify who received each test and why.
• Check order, interpreter knowledge, verification, flow and analysis.
• Keep comparative bias separate from applicability and practical choice.
• Explain the judgement in words before using a colour or label.

---

Define the Comparative Accuracy Question

Name the patients, index tests, target condition, reference approach, setting and intended role before appraising the comparison.

Practical reading move for Define the Comparative Accuracy Question

Start by drawing the intended clinical comparison: the same eligible patients receive both tests, or comparable groups are allocated fairly. Then mark where the actual study departs from that design. Under Define the Comparative Accuracy Question, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Define the Comparative Accuracy Question

In a fully paired study, every participant receives test A and test B, so patient characteristics are naturally shared; the reader still checks whether order, fatigue or knowledge of the first result could influence the second. Apply the example to define the comparative accuracy question by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

QUADAS-C assesses risk of bias in within-study comparisons of two or more index tests; it does not assess indirect comparisons assembled from separate single-test studies. For define the comparative accuracy question, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Use QUADAS-C Alongside QUADAS-3

The extension cannot stand alone; in 2026 the official guidance recommends adaptation alongside the current main QUADAS tool.

Practical reading move for Use QUADAS-C Alongside QUADAS-3

Build two columns for the individual test-accuracy appraisal and a third for the comparative question. A test can be appraised carefully on its own while the comparison between tests remains biased. Under Use QUADAS-C Alongside QUADAS-3, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Use QUADAS-C Alongside QUADAS-3

A randomised comparison can be fair only when the allocation sequence is genuinely random and concealed. Alternating assignment or allocation by clinic day is predictable and may create different patient groups. Apply the example to use quadas-c alongside quadas-3 by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

The official Bristol page says QUADAS-C was developed with QUADAS-2, cannot be used alone and now needs adaptation alongside QUADAS-3 using the QUADAS-3 explanation-and-elaboration guidance. For use quadas-c alongside quadas-3, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Identify Every Index Test and Version

Device model, assay, threshold, operator and interpretation rule must remain attached to each test.

Practical reading move for Identify Every Index Test and Version

Trace allocation, test order, interpreter knowledge and reference-standard verification as a sequence. Comparative bias often enters through an advantage given to one test rather than through an obviously defective test. Under Identify Every Index Test and Version, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Identify Every Index Test and Version

If experts interpret the new test while less experienced readers interpret the comparator, the study may measure expertise as well as test performance. The report should show whether that difference reflects practice or an unfair advantage. Apply the example to identify every index test and version by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

QUADAS-C does not assess applicability. Keep comparative risk of bias separate from whether the participants, tests and target condition match the review question. For identify every index test and version, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Choose the Correct Unit of Comparison

A within-study test comparison differs from an indirect contrast between separate single-test studies.

Practical reading move for Choose the Correct Unit of Comparison

Record whether the design is fully paired, randomised, partially paired or unpaired. The design label matters only when the report also explains how patients reached each test and which data entered the comparison. Under Choose the Correct Unit of Comparison, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Choose the Correct Unit of Comparison

When only positive results from one index test receive definitive verification, the apparent comparison can shift because target status is established differently. Follow both verification routes and every excluded result. Apply the example to choose the correct unit of comparison by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

A low-risk comparison does not prove that either test improves patient outcomes, is affordable or is the best choice for an individual. For choose the correct unit of comparison, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Classify the Comparative Study Design

Fully paired, randomised, partially paired and unpaired designs create different routes to comparability.

Practical reading move for Classify the Comparative Study Design

Use a comparison ledger with columns for participant spectrum, conduct, interpretation, verification, flow and analysis. Put the same clinical question above every column so the assessment does not drift. Under Classify the Comparative Study Design, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Classify the Comparative Study Design

A partially paired study may send a non-random subset to both tests. If the paired subset contains the most difficult cases, the comparison and the single-test estimates answer different population questions. Apply the example to classify the comparative study design by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

Incomplete reporting can make a judgement unclear, but unclear is not evidence that the study was badly conducted. It identifies an information gap that should remain visible. For classify the comparative study design, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Draw Participant and Test Flow

Show who received each test, in which order, with which reference assessment and who entered analysis.

Practical reading move for Draw Participant and Test Flow

Answer signalling questions with a source quotation or precise paraphrase and a mechanism. ‘No’ or ‘unclear’ becomes useful when the note says how the feature could favour one test. Under Draw Participant and Test Flow, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Draw Participant and Test Flow

A review comparing two screening tests should specify the same target condition, threshold role and intended setting. Otherwise a numerical contrast may combine tests serving different clinical purposes. Apply the example to draw participant and test flow by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

The extension supports systematic-review appraisal; it is not a checklist for awarding a universal winner between technologies. For draw participant and test flow, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Appraise Single-Test Accuracy First

The comparison builds on appraisal of each test’s accuracy estimate and should not erase individual limitations.

Practical reading move for Appraise Single-Test Accuracy First

Prepare the rationale before choosing the display colour. Name the domain, the design feature, the direction of possible advantage and the comparison that may be distorted. Under Appraise Single-Test Accuracy First, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Appraise Single-Test Accuracy First

A school science group can model the logic with two thermometers: test the same objects, randomise order, blind the reader to the other result and keep the reference measurement identical. Apply the example to appraise single-test accuracy first by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

Respect other languages and settings: translation, operator training and local pathways can affect implementation even when the comparative design is strong. For appraise single-test accuracy first, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Domain 1: Examine Patient Selection

Recruitment and exclusions determine whether test groups or paired participants form a fair clinical comparison.

Practical reading move for Domain 1: Examine Patient Selection

Start by drawing the intended clinical comparison: the same eligible patients receive both tests, or comparable groups are allocated fairly. Then mark where the actual study departs from that design. Under Domain 1: Examine Patient Selection, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Domain 1: Examine Patient Selection

In a fully paired study, every participant receives test A and test B, so patient characteristics are naturally shared; the reader still checks whether order, fatigue or knowledge of the first result could influence the second. Apply the example to domain 1: examine patient selection by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

QUADAS-C assesses risk of bias in within-study comparisons of two or more index tests; it does not assess indirect comparisons assembled from separate single-test studies. For domain 1: examine patient selection, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Prefer Fully Paired or Randomised Designs

These designs usually protect comparability better than non-random subsets or unrelated patient groups.

Practical reading move for Prefer Fully Paired or Randomised Designs

Build two columns for the individual test-accuracy appraisal and a third for the comparative question. A test can be appraised carefully on its own while the comparison between tests remains biased. Under Prefer Fully Paired or Randomised Designs, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Prefer Fully Paired or Randomised Designs

A randomised comparison can be fair only when the allocation sequence is genuinely random and concealed. Alternating assignment or allocation by clinic day is predictable and may create different patient groups. Apply the example to prefer fully paired or randomised designs by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

The official Bristol page says QUADAS-C was developed with QUADAS-2, cannot be used alone and now needs adaptation alongside QUADAS-3 using the QUADAS-3 explanation-and-elaboration guidance. For prefer fully paired or randomised designs, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Check Random Sequence Generation

A truly unpredictable sequence reduces systematic allocation of different patients to different tests.

Practical reading move for Check Random Sequence Generation

Trace allocation, test order, interpreter knowledge and reference-standard verification as a sequence. Comparative bias often enters through an advantage given to one test rather than through an obviously defective test. Under Check Random Sequence Generation, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Check Random Sequence Generation

If experts interpret the new test while less experienced readers interpret the comparator, the study may measure expertise as well as test performance. The report should show whether that difference reflects practice or an unfair advantage. Apply the example to check random sequence generation by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

QUADAS-C does not assess applicability. Keep comparative risk of bias separate from whether the participants, tests and target condition match the review question. For check random sequence generation, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Check Allocation Concealment

Concealment prevents recruiters from steering particular patients toward a preferred test.

Practical reading move for Check Allocation Concealment

Record whether the design is fully paired, randomised, partially paired or unpaired. The design label matters only when the report also explains how patients reached each test and which data entered the comparison. Under Check Allocation Concealment, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Check Allocation Concealment

When only positive results from one index test receive definitive verification, the apparent comparison can shift because target status is established differently. Follow both verification routes and every excluded result. Apply the example to check allocation concealment by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

A low-risk comparison does not prove that either test improves patient outcomes, is affordable or is the best choice for an individual. For check allocation concealment, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Understand Partial Pairing

A subset receiving both tests can strengthen or distort comparison depending on how that subset was chosen.

Practical reading move for Understand Partial Pairing

Use a comparison ledger with columns for participant spectrum, conduct, interpretation, verification, flow and analysis. Put the same clinical question above every column so the assessment does not drift. Under Understand Partial Pairing, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Understand Partial Pairing

A partially paired study may send a non-random subset to both tests. If the paired subset contains the most difficult cases, the comparison and the single-test estimates answer different population questions. Apply the example to understand partial pairing by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

Incomplete reporting can make a judgement unclear, but unclear is not evidence that the study was badly conducted. It identifies an information gap that should remain visible. For understand partial pairing, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Domain 2: Compare Test Conduct

Differences in setting, expertise, specimen handling or protocol can advantage one index test.

Practical reading move for Domain 2: Compare Test Conduct

Answer signalling questions with a source quotation or precise paraphrase and a mechanism. ‘No’ or ‘unclear’ becomes useful when the note says how the feature could favour one test. Under Domain 2: Compare Test Conduct, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Domain 2: Compare Test Conduct

A review comparing two screening tests should specify the same target condition, threshold role and intended setting. Otherwise a numerical contrast may combine tests serving different clinical purposes. Apply the example to domain 2: compare test conduct by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

The extension supports systematic-review appraisal; it is not a checklist for awarding a universal winner between technologies. For domain 2: compare test conduct, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Check Knowledge of the Other Test

Subjective interpretation may shift when the reader knows the competing test’s result.

Practical reading move for Check Knowledge of the Other Test

Prepare the rationale before choosing the display colour. Name the domain, the design feature, the direction of possible advantage and the comparison that may be distorted. Under Check Knowledge of the Other Test, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Check Knowledge of the Other Test

A school science group can model the logic with two thermometers: test the same objects, randomise order, blind the reader to the other result and keep the reference measurement identical. Apply the example to check knowledge of the other test by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

Respect other languages and settings: translation, operator training and local pathways can affect implementation even when the comparative design is strong. For check knowledge of the other test, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Check Order and Interference

The first test can alter the patient, specimen or reader experience available to the second.

Practical reading move for Check Order and Interference

Start by drawing the intended clinical comparison: the same eligible patients receive both tests, or comparable groups are allocated fairly. Then mark where the actual study departs from that design. Under Check Order and Interference, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Check Order and Interference

In a fully paired study, every participant receives test A and test B, so patient characteristics are naturally shared; the reader still checks whether order, fatigue or knowledge of the first result could influence the second. Apply the example to check order and interference by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

QUADAS-C assesses risk of bias in within-study comparisons of two or more index tests; it does not assess indirect comparisons assembled from separate single-test studies. For check order and interference, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Judge Whether One Test Was Favoured

Unequal training, thresholds, equipment or interpretation support may make the comparison unfair.

Practical reading move for Judge Whether One Test Was Favoured

Build two columns for the individual test-accuracy appraisal and a third for the comparative question. A test can be appraised carefully on its own while the comparison between tests remains biased. Under Judge Whether One Test Was Favoured, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Judge Whether One Test Was Favoured

A randomised comparison can be fair only when the allocation sequence is genuinely random and concealed. Alternating assignment or allocation by clinic day is predictable and may create different patient groups. Apply the example to judge whether one test was favoured by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

The official Bristol page says QUADAS-C was developed with QUADAS-2, cannot be used alone and now needs adaptation alongside QUADAS-3 using the QUADAS-3 explanation-and-elaboration guidance. For judge whether one test was favoured, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Domain 3: Keep the Reference Standard Comparable

Both index tests need target-condition classification that is sufficiently independent and consistently applied.

Practical reading move for Domain 3: Keep the Reference Standard Comparable

Trace allocation, test order, interpreter knowledge and reference-standard verification as a sequence. Comparative bias often enters through an advantage given to one test rather than through an obviously defective test. Under Domain 3: Keep the Reference Standard Comparable, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Domain 3: Keep the Reference Standard Comparable

If experts interpret the new test while less experienced readers interpret the comparator, the study may measure expertise as well as test performance. The report should show whether that difference reflects practice or an unfair advantage. Apply the example to domain 3: keep the reference standard comparable by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

QUADAS-C does not assess applicability. Keep comparative risk of bias separate from whether the participants, tests and target condition match the review question. For domain 3: keep the reference standard comparable, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Domain 4: Follow Timing and Verification

Timing, missing results and different verification routes can distort relative accuracy.

Practical reading move for Domain 4: Follow Timing and Verification

Record whether the design is fully paired, randomised, partially paired or unpaired. The design label matters only when the report also explains how patients reached each test and which data entered the comparison. Under Domain 4: Follow Timing and Verification, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Domain 4: Follow Timing and Verification

When only positive results from one index test receive definitive verification, the apparent comparison can shift because target status is established differently. Follow both verification routes and every excluded result. Apply the example to domain 4: follow timing and verification by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

A low-risk comparison does not prove that either test improves patient outcomes, is affordable or is the best choice for an individual. For domain 4: follow timing and verification, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Inspect the Comparative Analysis

The statistical analysis must respect paired or randomised data and the actual design.

Practical reading move for Inspect the Comparative Analysis

Use a comparison ledger with columns for participant spectrum, conduct, interpretation, verification, flow and analysis. Put the same clinical question above every column so the assessment does not drift. Under Inspect the Comparative Analysis, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Inspect the Comparative Analysis

A partially paired study may send a non-random subset to both tests. If the paired subset contains the most difficult cases, the comparison and the single-test estimates answer different population questions. Apply the example to inspect the comparative analysis by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

Incomplete reporting can make a judgement unclear, but unclear is not evidence that the study was badly conducted. It identifies an information gap that should remain visible. For inspect the comparative analysis, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Separate Risk of Bias From Applicability

QUADAS-C judges comparative bias and does not add an applicability judgement.

Practical reading move for Separate Risk of Bias From Applicability

Answer signalling questions with a source quotation or precise paraphrase and a mechanism. ‘No’ or ‘unclear’ becomes useful when the note says how the feature could favour one test. Under Separate Risk of Bias From Applicability, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Separate Risk of Bias From Applicability

A review comparing two screening tests should specify the same target condition, threshold role and intended setting. Otherwise a numerical contrast may combine tests serving different clinical purposes. Apply the example to separate risk of bias from applicability by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

The extension supports systematic-review appraisal; it is not a checklist for awarding a universal winner between technologies. For separate risk of bias from applicability, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Write the Domain Rationale

A reproducible note names the design feature, evidence location and possible direction of comparative distortion.

Practical reading move for Write the Domain Rationale

Prepare the rationale before choosing the display colour. Name the domain, the design feature, the direction of possible advantage and the comparison that may be distorted. Under Write the Domain Rationale, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Write the Domain Rationale

A school science group can model the logic with two thermometers: test the same objects, randomise order, blind the reader to the other result and keep the reference measurement identical. Apply the example to write the domain rationale by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

Respect other languages and settings: translation, operator training and local pathways can affect implementation even when the comparative design is strong. For write the domain rationale, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Did You Know? C Means Comparative

QUADAS-C was created specifically for primary studies comparing the accuracy of two or more index tests.

Practical reading move for Did You Know? C Means Comparative

Start by drawing the intended clinical comparison: the same eligible patients receive both tests, or comparable groups are allocated fairly. Then mark where the actual study departs from that design. Under Did You Know? C Means Comparative, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Did You Know? C Means Comparative

In a fully paired study, every participant receives test A and test B, so patient characteristics are naturally shared; the reader still checks whether order, fatigue or knowledge of the first result could influence the second. Apply the example to did you know? c means comparative by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

QUADAS-C assesses risk of bias in within-study comparisons of two or more index tests; it does not assess indirect comparisons assembled from separate single-test studies. For did you know? c means comparative, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Did You Know? Indirect Comparisons Are Outside Scope

The tool does not assess comparisons constructed between separate studies that each evaluated only one test.

Practical reading move for Did You Know? Indirect Comparisons Are Outside Scope

Build two columns for the individual test-accuracy appraisal and a third for the comparative question. A test can be appraised carefully on its own while the comparison between tests remains biased. Under Did You Know? Indirect Comparisons Are Outside Scope, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Did You Know? Indirect Comparisons Are Outside Scope

A randomised comparison can be fair only when the allocation sequence is genuinely random and concealed. Alternating assignment or allocation by clinic day is predictable and may create different patient groups. Apply the example to did you know? indirect comparisons are outside scope by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

The official Bristol page says QUADAS-C was developed with QUADAS-2, cannot be used alone and now needs adaptation alongside QUADAS-3 using the QUADAS-3 explanation-and-elaboration guidance. For did you know? indirect comparisons are outside scope, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Handle Unclear Reporting Honestly

Missing information supports an unclear judgement or a request for clarification, not automatic reassurance.

Practical reading move for Handle Unclear Reporting Honestly

Trace allocation, test order, interpreter knowledge and reference-standard verification as a sequence. Comparative bias often enters through an advantage given to one test rather than through an obviously defective test. Under Handle Unclear Reporting Honestly, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Handle Unclear Reporting Honestly

If experts interpret the new test while less experienced readers interpret the comparator, the study may measure expertise as well as test performance. The report should show whether that difference reflects practice or an unfair advantage. Apply the example to handle unclear reporting honestly by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

QUADAS-C does not assess applicability. Keep comparative risk of bias separate from whether the participants, tests and target condition match the review question. For handle unclear reporting honestly, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Explain Comparative Accuracy to Students

Use a fair-test analogy while preserving patient spectrum, thresholds, reference status and analysis.

Practical reading move for Explain Comparative Accuracy to Students

Record whether the design is fully paired, randomised, partially paired or unpaired. The design label matters only when the report also explains how patients reached each test and which data entered the comparison. Under Explain Comparative Accuracy to Students, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Explain Comparative Accuracy to Students

When only positive results from one index test receive definitive verification, the apparent comparison can shift because target status is established differently. Follow both verification routes and every excluded result. Apply the example to explain comparative accuracy to students by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

A low-risk comparison does not prove that either test improves patient outcomes, is affordable or is the best choice for an individual. For explain comparative accuracy to students, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Help Parents Read Test Headlines

Ask whether the tests were compared in the same people or fairly allocated groups before accepting a winner.

Practical reading move for Help Parents Read Test Headlines

Use a comparison ledger with columns for participant spectrum, conduct, interpretation, verification, flow and analysis. Put the same clinical question above every column so the assessment does not drift. Under Help Parents Read Test Headlines, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Help Parents Read Test Headlines

A partially paired study may send a non-random subset to both tests. If the paired subset contains the most difficult cases, the comparison and the single-test estimates answer different population questions. Apply the example to help parents read test headlines by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

Incomplete reporting can make a judgement unclear, but unclear is not evidence that the study was badly conducted. It identifies an information gap that should remain visible. For help parents read test headlines, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Connect Appraisal to Review Decisions

Sensitivity analyses can explore whether conclusions change when high-risk comparisons are excluded.

Practical reading move for Connect Appraisal to Review Decisions

Answer signalling questions with a source quotation or precise paraphrase and a mechanism. ‘No’ or ‘unclear’ becomes useful when the note says how the feature could favour one test. Under Connect Appraisal to Review Decisions, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Connect Appraisal to Review Decisions

A review comparing two screening tests should specify the same target condition, threshold role and intended setting. Otherwise a numerical contrast may combine tests serving different clinical purposes. Apply the example to connect appraisal to review decisions by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

The extension supports systematic-review appraisal; it is not a checklist for awarding a universal winner between technologies. For connect appraisal to review decisions, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Write a Responsible Comparison Summary

State which comparison is supported, the decisive limitation and why accuracy alone does not settle practical use.

Practical reading move for Write a Responsible Comparison Summary

Prepare the rationale before choosing the display colour. Name the domain, the design feature, the direction of possible advantage and the comparison that may be distorted. Under Write a Responsible Comparison Summary, write the operative noun, the comparison or synthesis it belongs to, the relevant source location and the exact reason the information changes interpretation.

Worked example for Write a Responsible Comparison Summary

A school science group can model the logic with two thermometers: test the same objects, randomise order, blind the reader to the other result and keep the reference measurement identical. Apply the example to write a responsible comparison summary by separating what the report states, what the reader can reasonably infer and what remains unknown.

Boundary, next step and transfer

Respect other languages and settings: translation, operator training and local pathways can affect implementation even when the comparative design is strong. For write a responsible comparison summary, end with a bounded sentence: what the evidence supports, which limitation matters most, and what document, analysis or local question would reduce uncertainty. Students can practise the same move in projects; parents can use it when reading evidence claims; researchers and professionals can use it to make handovers and decisions auditable.

---

Frequently Asked Questions

What does the C in QUADAS-C mean?

Comparative.

What does QUADAS-C assess?

Risk of bias in within-study comparisons of the accuracy of two or more index tests.

Can QUADAS-C be used alone?

No. The official page says it must be used alongside the main QUADAS tool.

How should it be used in 2026?

The official page recommends adapting it alongside QUADAS-3 using the QUADAS-3 explanation-and-elaboration guidance.

Does it assess indirect comparisons between separate studies?

No.

Does QUADAS-C assess applicability?

No. Applicability remains a separate consideration.

Why are fully paired designs useful?

The same participants receive both tests, reducing patient-spectrum differences, although order and interference still need checking.

Why does allocation concealment matter?

It reduces the opportunity to direct particular patients toward one test.

Is the more accurate test automatically the best choice?

No. Outcomes, feasibility, cost, burden and local use still matter.

How can students practise the logic?

Design a fair comparison in which the same objects or randomly allocated groups receive tests under equal conditions.

---

Useful Next Reading

• Use the current official QUADAS-C overview and resources.
• Read the official QUADAS-C tool.
• Use the current QUADAS-3 appraisal framework.
• Read diagnostic-accuracy reporting with STARD 2015.
• Practise critical reading of claims, evidence and bias.

Closing thought: QUADAS-C makes a comparative accuracy claim inspectable by asking whether both tests received a genuinely fair chance within the same clinical question.