# Diagnostic Accuracy Study Report: [Index test] for [target condition]

> Report template for a study that estimates how well a test (the index
> test) identifies a target condition against a reference standard. The
> sections follow the topics and order of STARD 2015 (30 essential items,
> BMJ 2015;351:h5527), with the participant flow diagram and the 2x2
> cross-tabulation. STARD item numbers appear in brackets, for example
> [12a]. Topics are written in original words; use the template with the
> official STARD 2015 checklist and flow diagram from the EQUATOR
> Network. For tests that use artificial intelligence, also use STARD-AI
> (2025). This is a reporting format, not clinical, statistical or
> regulatory advice. Replace every `[bracketed]` value and every row
> marked *Example*.

<!--
Guidance: How to use this template
- Fill the flow diagram counts (section 5.1) and the 2x2 table (section
  5.4) first. Every other number (prevalence, sensitivity, predictive
  values) is computed from them and must reconcile.
- Keep the bracketed STARD 2015 item numbers so you can complete the
  official checklist quickly (cross-reference in section 8.1).
- Describe the clinical pathway: where the test will be used, on whom,
  and what decision it informs. Accuracy figures without an intended use
  cannot be interpreted.
- Rows marked *Example* describe a fictional study of a point-of-care
  rapid antigen test (coded RAT-01) for group A streptococcal pharyngitis
  in children, run by the fictional Westmere Paediatric Network, with
  throat culture as the reference standard. All numbers are invented and
  were computed so that they add up. Delete them when you fill your own.
- Never include identifying participant data. Follow your ethics approval
  conditions and local privacy law.
-->

## Document control

<!-- Guidance: Treat the manuscript as a controlled document. Keep the protocol version and registration identical to what you report in section 7. -->

| Field | Value |
|---|---|
| Document ID | [DAS-001] |
| Study acronym | [ACRONYM] |
| Index test(s) | [Name, manufacturer, version or software release] |
| Reference standard | [Name and version] |
| Target condition | [Condition and definition] |
| Registration | [Registry name and number] |
| Protocol version reported against | [Protocol v2.0, YYYY-MM-DD] |
| Reporting guideline | STARD 2015 [plus STARD-AI or STARD for Abstracts where used] |
| Version | [0.1] |
| Status | [Draft / In review / Approved] |
| Owner | [@principal-investigator] |
| Approvers | [@statistician], [@laboratory-lead] |
| Last updated | YYYY-MM-DD |
| Classification | [Confidential until publication] |

### Approvals

| Role | Name | Decision | Date |
|---|---|---|---|
| Principal investigator | [@name] | [Approved] | YYYY-MM-DD |
| Statistician | [@name] | [Approved] | YYYY-MM-DD |
| Laboratory or reference standard lead | [@name] | [Approved] | YYYY-MM-DD |

### Revision history

| Version | Date | Author | Change summary |
|---|---|---|---|
| 0.1 | YYYY-MM-DD | [@name] | First draft after data lock |
| 0.2 | YYYY-MM-DD | [@name] | *Example:* Added sensitivity analysis for invalid index test results |
| 1.0 | YYYY-MM-DD | [@name] | Approved for submission |

## Contents

<!-- Guidance: Keep the numbering in step with the headings. -->

1. [Title and abstract](#1-title-and-abstract)
2. [Introduction](#2-introduction)
3. [Methods](#3-methods)
4. [Analysis](#4-analysis)
5. [Results](#5-results)
6. [Discussion](#6-discussion)
7. [Other information](#7-other-information)
8. [STARD cross-reference, glossary and references](#8-stard-cross-reference-glossary-and-references)

## 1. Title and abstract

<!-- Guidance: Name the study as a diagnostic accuracy study and report at least one accuracy measure in the title or abstract, for example sensitivity, specificity, predictive values or the area under the ROC curve. Use STARD for Abstracts for the abstract. -->

The title and abstract let readers recognise the study design and its
main result.

### 1.1 Title [1]

**Diagnostic accuracy of [index test] for [target condition] in
[population and setting]: a [prospective, cross-sectional] study**

*Example:* "Sensitivity and specificity of a point-of-care rapid antigen
test for group A streptococcal pharyngitis in children: a prospective
diagnostic accuracy study."

### 1.2 Structured abstract [2]

| Heading | Content |
|---|---|
| Objective | [Accuracy of which test, for which condition, in whom] |
| Design and setting | [Prospective, consecutive series, number and type of sites] |
| Participants | [Eligibility in brief] |
| Index test and reference standard | [Both, with positivity thresholds] |
| Results | *Example:* Of 640 children enrolled, 620 were included in the analysis; 170 (27.4%) had a positive culture. Sensitivity 87.1% (95% CI 81.2 to 91.3), specificity 93.1% (90.4 to 95.1). |
| Conclusions | [Intended use and limits, in one sentence] |
| Registration | [Registry and number] |

## 2. Introduction

<!-- Guidance: Describe the clinical pathway and the role the index test should play in it (replacement, triage or add-on). The required accuracy depends on that role. -->

The introduction sets out why the test matters and what the study tries
to estimate.

### 2.1 Scientific and clinical background [3]

[Describe the target condition, the current pathway and the intended use
and clinical role of the index test. Say what decision a positive or
negative result should change.]

*Example:* Sore throat is common in children and most cases are viral.
Culture takes 24 to 48 hours. A rapid test at the first visit could act
as a replacement for culture in deciding on antibiotics, so a high
sensitivity is needed to avoid missed infections.

| Role of the index test | Meaning | Accuracy usually emphasised |
|---|---|---|
| Replacement | Replaces an existing test | Similar or better than the existing test overall |
| Triage | Used before the existing test; only positives (or negatives) go on | High sensitivity (rule-out) or high specificity (rule-in) |
| Add-on | Used after the existing test | Gain in accuracy where the existing test is uncertain |

### 2.2 Objectives and hypotheses [4]

| ID | Objective | Hypothesis or target |
|---|---|---|
| OBJ-01 *(Example)* | Estimate the sensitivity and specificity of RAT-01 against throat culture in children aged 3 to 14 with sore throat | Lower 95% confidence limit of sensitivity at least 80% |
| OBJ-02 *(Example)* | Estimate positive and negative predictive values in this setting | Descriptive |
| OBJ-03 | [Variability by age group or symptom score] | [Prespecified or exploratory] |

## 3. Methods

<!-- Guidance: Give enough detail for another team to repeat the study. The risk of bias in accuracy studies comes mostly from how participants were selected, how the tests were done and read, and who was left out. -->

The methods describe the design, participants and tests as they were
used.

### 3.1 Study design [5]

[State whether data collection was planned before the index test and
reference standard were done (prospective) or after (retrospective).]

### 3.2 Participants [6, 7, 8, 9]

<!-- Guidance: Report the eligibility criteria, how potentially eligible people were identified (by symptoms, by earlier test results, from a registry), the setting, locations and dates, and whether the series was consecutive, random or convenience. Case-control sampling (people with known disease plus healthy controls) inflates accuracy; say so if you used it. -->

| Item | Detail |
|---|---|
| Eligibility criteria [6] | *Example:* Children aged 3 to 14 presenting with sore throat of 7 days or less; excluded if antibiotics in the previous 14 days |
| Basis for identification [7] | *Example:* Presenting symptoms at triage |
| Setting, locations, dates [8] | *Example:* 6 urgent care clinics of the Westmere Paediatric Network, YYYY-MM-DD to YYYY-MM-DD |
| Sampling [9] | *Example:* Consecutive eligible children during clinic opening hours |

### 3.3 Test methods [10a, 10b, 11, 12a, 12b, 13a, 13b]

<!-- Guidance: Describe both tests in enough detail to replicate them, explain why the reference standard was chosen, define what counted as positive (with a note on whether the threshold was prespecified), and say what information the people performing or reading each test had. -->

| Item | Index test | Reference standard |
|---|---|---|
| Description [10a, 10b] | *Example:* RAT-01 lateral flow assay, lot numbers recorded, performed by triage nurses after a 30-minute training | Throat culture on blood agar, read at 24 and 48 hours by laboratory staff |
| Rationale for the reference standard [11] | Not applicable | [Why it is the best available method; known limitations] |
| Positivity definition and threshold [12a, 12b] | *Example:* Any visible test line within 10 minutes, as in the instructions for use; prespecified | Any growth of group A streptococcus; prespecified |
| Information available to readers [13a, 13b] | *Example:* Readers saw symptoms but not culture results | Laboratory staff were not told the rapid test result |

```mermaid
flowchart LR
    A["Eligible child with sore throat"] --> B["Two throat swabs taken at one visit"]
    B --> C["Swab 1: index test RAT-01 at point of care"]
    B --> D["Swab 2: reference standard culture in laboratory"]
    C --> E["Result recorded before culture result known"]
    D --> F["Culture result recorded without index result"]
    E --> G["Cross-tabulation, 2x2 table"]
    F --> G
```

## 4. Analysis

<!-- Guidance: State how accuracy was estimated and compared, how indeterminate and missing results were handled, which variability analyses were prespecified, and how the sample size was set. -->

This section describes the statistical methods.

### 4.1 Estimating accuracy [14]

Accuracy is derived from the 2x2 table of index test result against
reference standard result:

| | Reference positive | Reference negative |
|---|---|---|
| Index positive | TP (true positive) | FP (false positive) |
| Index negative | FN (false negative) | TN (true negative) |

$$
\text{Sensitivity} = \frac{TP}{TP + FN}, \qquad \text{Specificity} = \frac{TN}{TN + FP}
$$

$$
\text{LR}^{+} = \frac{\text{Sensitivity}}{1 - \text{Specificity}}, \qquad \text{LR}^{-} = \frac{1 - \text{Sensitivity}}{\text{Specificity}}
$$

[Name the method for confidence intervals, for example Wilson score
intervals for proportions and the log method for likelihood ratios, and
the software and version.] Predictive values depend on prevalence; report
them only for the setting studied, or recalculate them for other
prevalences with the likelihood ratios.

### 4.2 Indeterminate and missing results [15, 16]

| Situation | Handling |
|---|---|
| Invalid or indeterminate index test [15] | *Example:* Reported separately; excluded from the main analysis; sensitivity analysis counting them as wrong (worst case) and as right (best case) |
| Inconclusive reference standard [15] | [Repeat test / exclude / adjudication panel] |
| Missing index test or reference standard [16] | *Example:* Reported in the flow diagram; no imputation in the main analysis |

### 4.3 Variability analyses [17]

| ID | Analysis | Prespecified? |
|---|---|---|
| VAR-01 *(Example)* | Sensitivity by age group (3 to 5, 6 to 14) | Yes |
| VAR-02 | [By site or by reader] | [Yes / No, exploratory] |

### 4.4 Sample size [18]

<!-- Guidance: Base the sample size on the precision you want for the key accuracy measure, inflated for the expected prevalence and for losses. Cite the source of each assumption. -->

The number of participants with the target condition needed for a
given precision of sensitivity is:

$$
n_{\text{cases}} = \frac{z_{1-\alpha/2}^{2}\ \text{Se}\,(1 - \text{Se})}{d^{2}}, \qquad n_{\text{total}} = \frac{n_{\text{cases}}}{\text{prevalence}}
$$

*Example:* expected sensitivity $\text{Se} = 0.85$, half-width
$d = 0.06$ and $z = 1.96$ give
$n_{\text{cases}} = 3.8416 \times 0.1275 / 0.0036 = 136.05$, rounded up
to 137 children with a positive culture. With an expected prevalence of
25%, $137 / 0.25 = 548$ children; allowing 10% for missing or invalid
results, $548 / 0.9 = 608.9$, rounded up to 609. Recruitment continued
to the end of the planned season and enrolled 640.

## 5. Results

<!-- Guidance: Report participants first, then the cross-tabulation, then the estimates. Readers check that the flow diagram, the 2x2 table and the estimates use the same numbers. -->

This section reports who took part and how accurate the test was.

### 5.1 Participant flow [19]

<!-- Guidance: Show how many people were assessed, excluded with reasons, received the index test, the reference standard, and how many ended up in each cell of the 2x2 table. Replace this sketch with the official STARD 2015 flow diagram in the final manuscript. -->

The diagram follows participants from eligibility to the final analysis.

```mermaid
flowchart TD
    A["Children assessed for eligibility (n = 812)"] --> B["Excluded (n = 172)"]
    A --> C["Enrolled (n = 640)"]
    C --> D["Index test not done, child refused swab (n = 5)"]
    C --> E["Index test done (n = 635)"]
    E --> F["Positive (n = 184)"]
    E --> G["Negative (n = 442)"]
    E --> H["Invalid result (n = 9)"]
    F --> I["No reference standard (n = 5)"]
    F --> J["Reference standard done (n = 179): culture positive 148, negative 31"]
    G --> K["No reference standard (n = 1)"]
    G --> L["Reference standard done (n = 441): culture positive 22, negative 419"]
    H --> M["Reference standard done (n = 9): culture positive 3, negative 6"]
```

*Example:* reasons for exclusion before enrolment: did not meet the
criteria 118, declined 41, other 13 (total 172). Children in the main
analysis: $640 - 5 - 5 - 1 - 9 = 620$.

### 5.2 Baseline characteristics [20, 21a, 21b]

<!-- Guidance: Describe participants, the severity of disease in those with the target condition and the alternative diagnoses in those without it. Spectrum strongly affects sensitivity and specificity. -->

| Characteristic | Culture positive (n = [170]) | Culture negative (n = [450]) |
|---|---|---|
| Age, years, median (IQR) | *Example:* 7 (5 to 10) | 6 (4 to 9) |
| Female, n (%) | [84 (49.4)] | [226 (50.2)] |
| Clinical score at presentation, median | [Value] | [Value] |
| Severity in those with the condition [21a] | [Distribution, for example by clinical score] | Not applicable |
| Alternative diagnoses [21b] | Not applicable | [Viral pharyngitis, other] |

### 5.3 Time interval and interventions [22]

[Time between index test and reference standard, and any treatment
started in between.] *Example:* both swabs were taken at the same visit
(interval 0 days); no antibiotic was given before swabbing.

### 5.4 Cross-tabulation [23]

<!-- Guidance: Give the full 2x2 table (or the distribution of continuous results by reference standard status). Without it readers cannot recompute your estimates. -->

| | Culture positive | Culture negative | Total |
|---|---|---|---|
| RAT-01 positive *(Example)* | 148 (TP) | 31 (FP) | 179 |
| RAT-01 negative *(Example)* | 22 (FN) | 419 (TN) | 441 |
| Total | 170 | 450 | 620 |

### 5.5 Estimates of accuracy [24]

| ID | Measure | Calculation | Estimate (95% CI) |
|---|---|---|---|
| EST-01 *(Example)* | Sensitivity | 148 / 170 | 87.1% (81.2 to 91.3) |
| EST-02 *(Example)* | Specificity | 419 / 450 | 93.1% (90.4 to 95.1) |
| EST-03 *(Example)* | Positive predictive value | 148 / 179 | 82.7% (76.5 to 87.5) |
| EST-04 *(Example)* | Negative predictive value | 419 / 441 | 95.0% (92.6 to 96.7) |
| EST-05 *(Example)* | Positive likelihood ratio | 0.871 / 0.069 | 12.64 (8.95 to 17.84) |
| EST-06 *(Example)* | Negative likelihood ratio | 0.129 / 0.931 | 0.14 (0.09 to 0.21) |
| EST-07 *(Example)* | Prevalence in the sample | 170 / 620 | 27.4% (24.1 to 31.1) |

Confidence intervals: Wilson score for proportions, log method for
likelihood ratios. The ratios in the calculation column are rounded for
display; the estimates use unrounded values.

| ID | Sensitivity analysis | Sensitivity | Specificity |
|---|---|---|---|
| SEN-01 *(Example)* | 9 invalid results counted as wrong (worst case) | 148 / 173 = 85.5% | 419 / 456 = 91.9% |
| SEN-02 *(Example)* | 9 invalid results counted as right (best case) | 151 / 173 = 87.3% | 425 / 456 = 93.2% |
| VAR-01 | [Results by age group from section 4.3] | [Value] | [Value] |

### 5.6 Adverse events [25]

[Any adverse events from performing the index test or the reference
standard.] *Example:* no adverse events beyond brief gagging during
swabbing were recorded.

## 6. Discussion

<!-- Guidance: Separate the limitations (bias, imprecision, applicability) from the implications, and link both to the intended role of the test from section 2.1. -->

The discussion interprets the estimates for the intended use of the test.

### 6.1 Limitations [26]

- [Sources of bias, for example imperfect reference standard or partial verification]
- [Statistical uncertainty, for example a wide interval for sensitivity in a subgroup]
- [Generalisability: setting, prevalence, operators and test version]

### 6.2 Implications for practice [27]

[State what the results mean for the intended use and clinical role.]
*Example:* with a negative likelihood ratio of 0.14, a negative result
lowers the probability of a positive culture from 27.4% to about 5.0%
in this setting. Whether that is low enough to withhold culture is a
clinical and guideline decision, not something this study settles.

## 7. Other information

<!-- Guidance: Registration, protocol access and funding are items 28 to 30. Declare any link with the test manufacturer. -->

| Item | Detail |
|---|---|
| Registration [28] | [Registry name and number] |
| Protocol access [29] | [Where the full protocol can be accessed] |
| Funding and role of funders [30] | [Funder; whether the manufacturer supplied tests, and its role in design, analysis and reporting] |
| Data and code | [Repository, access conditions] |

## 8. STARD cross-reference, glossary and references

<!-- Guidance: Complete the official STARD 2015 checklist with the page or section of each item and submit it with the manuscript. -->

This section maps the STARD 2015 items to this report and lists terms
and sources.

### 8.1 STARD 2015 cross-reference

| Item | Topic (paraphrased) | Section |
|---|---|---|
| 1, 2 | Title, abstract | 1.1, 1.2 |
| 3, 4 | Background and intended use, objectives | 2.1, 2.2 |
| 5 | Prospective or retrospective design | 3.1 |
| 6 to 9 | Participants: eligibility, identification, setting, sampling | 3.2 |
| 10a, 10b, 11 | Index test, reference standard, rationale | 3.3 |
| 12a, 12b | Positivity definitions and thresholds | 3.3 |
| 13a, 13b | Information available to test readers | 3.3 |
| 14, 15, 16 | Estimation methods, indeterminate and missing results | 4.1, 4.2 |
| 17, 18 | Variability analyses, sample size | 4.3, 4.4 |
| 19 | Participant flow diagram | 5.1 |
| 20, 21a, 21b | Baseline characteristics, severity, alternative diagnoses | 5.2 |
| 22 | Time interval and interventions between tests | 5.3 |
| 23 | Cross-tabulation | 5.4 |
| 24 | Accuracy estimates with precision | 5.5 |
| 25 | Adverse events from testing | 5.6 |
| 26, 27 | Limitations, implications | 6.1, 6.2 |
| 28, 29, 30 | Registration, protocol, funding | 7 |

### 8.2 Glossary

| Term | Meaning |
|---|---|
| Index test | The test whose accuracy is evaluated |
| Reference standard | The best available method for establishing whether the target condition is present |
| Likelihood ratio | How much a result changes the odds of the target condition |
| Partial verification | Only some participants receive the reference standard, often chosen by index test result |
| Spectrum | The mix of severity and alternative diagnoses in the participants |
| QUADAS-3 | Tool for assessing risk of bias and applicability of diagnostic accuracy studies in systematic reviews |

### 8.3 References and related templates

| Reference | Version or date | Use |
|---|---|---|
| STARD 2015, Bossuyt PM et al., BMJ 2015;351:h5527, https://doi.org/10.1136/bmj.h5527 | 2015 | Reporting items |
| STARD 2015 explanation and elaboration, Cohen JF et al., BMJ Open 2016;6:e012799, https://doi.org/10.1136/bmjopen-2016-012799 | 2016 | Detailed guidance |
| STARD checklist and flow diagram, https://www.equator-network.org/reporting-guidelines/stard/ | Checked 2026-10-30 | Submission files |
| STARD-AI, Sounderajah V et al., Nat Med 2025;31:3283-3289, https://doi.org/10.1038/s41591-025-03953-8 | 2025 | Tests that use artificial intelligence |
| STARD for Abstracts, Cohen JF et al., BMJ 2017;358:j3751 | 2017 | Abstract |
| QUADAS-3, Whiting P et al., Ann Intern Med 2026;179(4):548-555, https://doi.org/10.7326/ANNALS-25-02104 | 2026 | How reviewers will appraise the study |
| academic/stats-contingency-tables, academic/stats-confidence-intervals | This collection | 2x2 tables and intervals |
| academic/stats-power-sample-size | This collection | Sample size reasoning |
| science/observational-study, science/prediction-model-report | This collection | Related designs |
| science/systematic-review-report | This collection | Reviews that pool accuracy studies |

## Standard and licence note

Structure aligned with STARD 2015: Bossuyt PM, Reitsma JB, Bruns DE,
Gatsonis CA, Glasziou PP, Irwig L, et al., *STARD 2015: an updated list
of essential items for reporting diagnostic accuracy studies*, BMJ
2015;351:h5527 (https://doi.org/10.1136/bmj.h5527), listed as the
current guideline by the EQUATOR Network on 2026-10-30. This template
follows the order of its 30 items, paraphrased in original text, and
reproduces no checklist wording; use the official checklist and flow
diagram for submission and check the EQUATOR page for updates and
extensions. This is a reporting format, not clinical, statistical or
regulatory advice. Template licence: MIT.