STARD

STARD
Name	STARD
Full name	Standards for Reporting Diagnostic Accuracy Studies
First published	2003
Developer	STARD Group
Purpose	Improve reporting of diagnostic accuracy studies
Latest version	2015 (STARD 2015)

STARD STARD is a reporting guideline aimed at improving the quality, transparency, and completeness of published diagnostic accuracy studies. It provides an evidence-based checklist and flow diagram intended to reduce bias and enhance reproducibility in articles describing index tests and reference standards. STARD has been adopted and endorsed by journals, research institutions, and guideline developers to harmonize reporting across disciplines and specialties.

Background and purpose

STARD was initiated in response to concerns about incomplete reporting in diagnostic accuracy literature identified by researchers associated with BMJ, Cochrane Collaboration, QUADAS, and other methodological initiatives. The project was informed by empirical evaluations conducted by teams at University of Oxford, McMaster University, and University of Amsterdam, and by methodological frameworks advanced by CONSORT and PRISMA. The primary purpose is to enable readers—including clinicians from Mayo Clinic, Johns Hopkins Hospital, and Cleveland Clinic—systematic reviewers from Cochrane Diagnostic Test Accuracy Group, guideline developers at National Institute for Health and Care Excellence, and health technology assessors at IQWiG to judge potential biases, applicability, and reproducibility. By standardizing reporting, STARD seeks to facilitate uptake of evidence in clinical policy decisions at organizations such as World Health Organization and Centers for Disease Control and Prevention.

Development and editions

The original STARD statement was published in 2003 following a consensus meeting that included representatives from BMJ, Journal of the American Medical Association, and methodologists from University College London and Karolinska Institutet. An update, STARD 2015, was promulgated after a multi-stakeholder Delphi process involving editors from Lancet, Annals of Internal Medicine, and Radiology, as well as methodologists from Yale University and Imperial College London. The revision incorporated feedback from stakeholders including journal editors, peer reviewers, statisticians at Statistical Society of Canada, and clinical specialists from Royal College of Physicians. Guideline development used established methods from the Equator Network and aligned with reporting initiatives like CONSORT and TRIPOD.

Checklist items and structure

STARD 2015 comprises a core checklist of items and a flow diagram. The checklist specifies key elements that studies should report: study design, participant recruitment at sites such as Mayo Clinic Hospital or Mount Sinai Hospital, index tests (examples from Magnetic Resonance Imaging and Polymerase Chain Reaction assays used at CDC), reference standards (e.g., histopathology from Johns Hopkins Hospital), masking and blinding procedures, sample size calculations, and statistical methods for estimates such as sensitivity and specificity. The flow diagram mirrors participant flow conventions used by CONSORT and documents numbers of participants at each stage, exclusions, and reasons—information that benefits systematic reviewers at Cochrane and methodologists at Harvard T.H. Chan School of Public Health. The checklist items address sources of bias exemplified in classic studies from New England Journal of Medicine and analytical strategies discussed in textbooks from Cambridge University Press.

Use in research and reporting

Investigators from institutions like Stanford University, University of Toronto, and King's College London are recommended to use STARD during study planning, manuscript drafting, and peer review. Journals such as BMJ, PLoS Medicine, and The Lancet encourage or mandate STARD adherence in author instructions and submission checklists. Systematic reviewers at Cochrane and guideline panels at National Institutes of Health use STARD items to assess completeness of reporting and to inform risk-of-bias tools. Biostatisticians at Washington University in St. Louis and diagnostic laboratory directors at Mayo Clinic apply STARD items to evaluate reproducibility of diagnostic thresholds and to assess applicability to settings like primary care clinics and specialized centers such as Dana-Farber Cancer Institute.

Impact and evaluations

Empirical evaluations by teams at University of Oxford and McMaster University have shown that adoption of STARD is associated with improvements in reporting completeness, particularly for items such as participant flow and test execution. Journal endorsement by publishers including BMJ Publishing Group and Elsevier led to measurable increases in checklist adherence in some specialties, though gaps persist in reporting of sample size justification and handling of indeterminate results. Meta-researchers at ETH Zurich and University of Melbourne have identified variability in impact across fields such as radiology, laboratory medicine, and infectious disease diagnostics, echoing findings from studies published in JAMA and Annals of Internal Medicine.

Implementation and extensions

STARD has inspired context-specific extensions and tools: adaptations for imaging studies aligned with work from Radiological Society of North America, extensions for point-of-care tests informed by World Health Organization guidance, and integration with reporting platforms used by Open Science Framework. Training materials and explanatory documents developed by groups at University of Oxford and University of Amsterdam support implementation in low-resource settings and among research networks such as Global Health Network. Ongoing work links STARD with reporting standards like PRISMA-DTA for diagnostic test accuracy systematic reviews and with initiatives promoting data sharing from institutions including Wellcome Trust and European Commission.

Category:Reporting guidelines