Contents
Academy · Checklists · STARD 2015 checklist
STARD 2015 Gives Blinding Its Own Two Items, Because Unblinded Reads Are the Commonest Source of Inflated Accuracy
STARD 2015 is the 30-item standard for reporting diagnostic accuracy studies — how well an index test identifies a target condition against a reference standard. It is not for treatment-effect, prognostic or prediction-model studies.
STARD stands for Standards for Reporting of Diagnostic Accuracy Studies. STARD 2015, by Patrick Bossuyt, Johannes Reitsma and 14 co-authors for the STARD Group, updates the original 2003 statement and was published simultaneously in BMJ, Radiology and Clinical Chemistry in 2015.
What it is
Bossuyt PM, Reitsma JB, Bruns DE, et al., for the STARD Group. “STARD 2015: An Updated List of Essential Items for Reporting Diagnostic Accuracy Studies.” BMJ 2015;351:h5527. Open access.
Methodological reviews found diagnostic accuracy studies routinely missing key detail — how the reference standard was applied, whether readers were blinded, how indeterminate results were handled — making it hard to judge whether a reported sensitivity or specificity is trustworthy. STARD 2015 consolidates and updates the 2003 items to close that gap.
Diagnostic accuracy studies estimating sensitivity, specificity or predictive values for one or more tests against a reference standard — laboratory tests, imaging tests, or diagnostic clinical prediction rules.
Is this the right tool?
Use it when
Authors of a diagnostic accuracy manuscript, from drafting through peer review. Many radiology, pathology and laboratory-medicine journals require a completed STARD checklist at submission.
Not the right tool when
A treatment-effect study should use CONSORT, even if it involves a device. A prognostic study, or one developing/validating a multivariable prediction model, should generally use TRIPOD instead. STARD assumes a direct index-test-versus-reference-standard comparison in the same population — a purely descriptive test-characteristics study with no reference standard is out of scope.
The checklist
30 items under six section headers.
Title/abstract and introduction
| Item | What it requires |
|---|---|
| 1–2. Title and abstract | Identify the report as a diagnostic accuracy study with at least one accuracy measure; structured summary. |
| 3–4. Background and objectives | Scientific/clinical context and the index test’s intended role; study objectives. |
Methods: design and participants
| Item | What it requires |
|---|---|
| 5. Data collection | Prospective or retrospective, relative to the index test and reference standard. |
| 6–9. Participants | Eligibility; how potentially eligible participants were identified; setting/dates; consecutive, random or convenience sampling. |
Methods: test methods and analysis
| Item | What it requires |
|---|---|
| 10–11. Index test and reference standard | Both described in enough detail to replicate; the reference standard’s choice justified. |
| 12–13. Thresholds and blinding | Positivity cut-offs, pre-specified vs exploratory; whether index-test and reference-standard readers were blinded to each other and to other clinical information. |
| 14–18. Analysis | Accuracy estimation methods; indeterminate-result and missing-data handling; variability analyses; sample-size determination. |
Results and discussion
| Item | What it requires |
|---|---|
| 19–22. Participants | Flow diagram; baseline characteristics; disease-severity and alternative-diagnosis distribution; time interval between tests. |
| 23–25. Test results | Cross-tabulation of index vs reference results; accuracy estimates with confidence intervals; adverse events. |
| 26–30. Discussion and other information | Limitations and generalisability; clinical implications; registration; protocol access; funding. |
Applying it, with the evidence
A review of 300 diagnostic-accuracy studies in cardiac MRI, CT and SPECT (split into pre- and post-STARD periods and scored against the checklist) found adherence rose meaningfully for MRI and CT after the guideline existed, but stayed essentially flat for SPECT — showing a published checklist does not automatically raise reporting quality across every subfield. Journal impact factor predicted reporting quality more strongly than whether a journal’s author instructions even mentioned STARD. For diagnostic-accuracy research, incomplete reporting (missing patient flow, reference-standard timing, how indeterminate results were handled) directly threatens whether a test’s reported sensitivity and specificity would hold up in another population.
STARD applies wherever a study estimates a test’s diagnostic accuracy against a reference standard: radiology and medical imaging (MRI, CT, SPECT, ultrasound), laboratory medicine and clinical chemistry, cardiology, infectious disease rapid-test evaluation, and increasingly AI/machine-learning diagnostic model evaluation, reflected in the newer STARD-AI extension.
In order
Where it goes right
With it — no verifiable example found
Our research for this page did not turn up a named, checkable published paper that explicitly credits this checklist for a well-reported study. Rather than invent one, we are saying so. If you know of one, the EQUATOR Network page for this guideline is the place such examples tend to get collected.
Common mistakes
- Unreported or absent blinding is the most consistently documented source of inflated accuracy estimates. Items 13a/13b exist specifically to surface this.
- Accuracy figures without confidence intervals (item 24) are a frequent, avoidable omission. They limit how precisely a reader can judge the estimate.
- Indeterminate and missing results are sometimes excluded without disclosure. Items 15–16 call for this to be stated explicitly, since silent exclusion can bias the estimate.
Where it falls short
Without it — Maclean et al. 2013, 300 cardiac imaging studies
Scoring 300 diagnostic-accuracy studies in cardiac MRI, CT and SPECT (50 each, pre- and post-STARD) against the checklist, adherence rose from 65.3% to 74.1% for MRI and from 61.6% to 79.0% for CT, but barely moved for SPECT (71.9% to 71.5%, p=0.92) — no measurable improvement at all in that modality nine years after the guideline’s introduction. This audit used the original 25-criterion STARD list that preceded the 2015 update; it is cited here as the clearest available evidence of how unevenly a STARD checklist gets taken up across otherwise-similar imaging fields.
Maclean EN, Stone IS, Ceelen F, et al. “Quality of reporting in cardiac MRI, CT and SPECT diagnostic accuracy studies: Analysis of the impact of STARD criteria.” Journal of Cardiovascular Magnetic Resonance, 2013;15(Suppl 1):E76. PMC3559352.
Common questions
Not directly. STARD assumes a reference-standard comparison; a head-to-head comparison without one is a different design and should flag that limitation clearly.
STARD is for a single test (or a small number) evaluated against a reference standard; TRIPOD is for developing or validating a multivariable prediction model.
As far as this research could confirm, yes — no EQUATOR-listed update beyond 2015 was found. Check the EQUATOR STARD page directly before relying on this.
References and sources
- Bossuyt PM, Reitsma JB, Bruns DE, et al., for the STARD Group. STARD 2015: An Updated List of Essential Items for Reporting Diagnostic Accuracy Studies. BMJ 2015;351:h5527. read 8 October 2026. doi.org/10.1136/bmj.h5527
- EQUATOR Network. STARD 2015 checklist. read 8 October 2026. equator-network.org
Each source carries the date it was last read. Where a standard moves, this article is revised rather than left up with a note.
How to cite this article
Pick your referencing style and copy it. Your university or target journal may vary the punctuation — check its guide if it has one.
Pubrica Academy. (2026, October 8). STARD 2015 Gives Blinding Its Own Two Items, Because Unblinded Reads Are the Commonest Source of Inflated Accuracy. Pubrica. https://pubrica.com/insights/checklist/stard-2015-reporting-checklist/
Pubrica Academy. “STARD 2015 Gives Blinding Its Own Two Items, Because Unblinded Reads Are the Commonest Source of Inflated Accuracy.” Pubrica, 8 Oct. 2026, pubrica.com/insights/checklist/stard-2015-reporting-checklist/.
Pubrica Academy (2026) STARD 2015 Gives Blinding Its Own Two Items, Because Unblinded Reads Are the Commonest Source of Inflated Accuracy. Pubrica. Available at: https://pubrica.com/insights/checklist/stard-2015-reporting-checklist/ (Accessed: 15 October 2026).
Pubrica Academy. “STARD 2015 Gives Blinding Its Own Two Items, Because Unblinded Reads Are the Commonest Source of Inflated Accuracy.” Pubrica. October 8, 2026. https://pubrica.com/insights/checklist/stard-2015-reporting-checklist/
Pubrica Academy, ‘STARD 2015 Gives Blinding Its Own Two Items, Because Unblinded Reads Are the Commonest Source of Inflated Accuracy’ (Pubrica, 8 October 2026) <https://pubrica.com/insights/checklist/stard-2015-reporting-checklist/> accessed 15 October 2026.
Pubrica Academy. STARD 2015 Gives Blinding Its Own Two Items, Because Unblinded Reads Are the Commonest Source of Inflated Accuracy. [Internet]. Pubrica; 2026 Oct 8 [cited 2026 Oct 15]. Available from: https://pubrica.com/insights/checklist/stard-2015-reporting-checklist/
Pubrica Academy, “STARD 2015 Gives Blinding Its Own Two Items, Because Unblinded Reads Are the Commonest Source of Inflated Accuracy,” Pubrica, Oct. 8, 2026. [Online]. Available: https://pubrica.com/insights/checklist/stard-2015-reporting-checklist/
@misc{stard_2015_reporting_checklist_2026,
author = {{Pubrica Academy}},
title = {STARD 2015 Gives Blinding Its Own Two Items, Because Unblinded Reads Are the Commonest Source of Inflated Accuracy},
howpublished = {Pubrica},
year = {2026},
month = {oct},
url = {https://pubrica.com/insights/checklist/stard-2015-reporting-checklist/},
note = {Accessed: <span data-cite-acc="iso">2026-10-15</span>}
}This article is written to be learned from. Cite it if you are referring to what it says; if you are using the evidence behind it, cite the sources listed above instead.
Pubrica AcademyWritten and maintained by the subject matter experts of Pubrica who do this work. Articles are revised when the standard underneath them moves, and each one carries the date it was last checked.
How Pubrica helps
Pubrica’s biostatistics team checks blinding and reference-standard justification against the actual study design before drafting, since reviewers scrutinise these items first.