This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record.
1. Trial at a Glance
CHEST Extension was a completed single-group diagnostic study evaluating the Rapid Rhythm handheld 8-lead ECG device in patients with atrial fibrillation and heart murmurs. The registry reports an enrollment of 271 participants and two primary endpoints focused on how often Rapid Rhythm ECGs matched normal or abnormal findings from control ECGs.
| Feature | CHEST Extension |
|---|---|
| Study name | CHEST Extension |
| Official title | Wireless Handheld 8-lead ECG Device Performance in Heart Patients (Rapid Rhythm) |
| Phase | NA |
| Status | COMPLETED |
| Primary purpose | DIAGNOSTIC |
| Design model | SINGLE_GROUP |
| Allocation | NA |
| Masking | NONE |
| Enrollment | 271 |
| Number of arms | 1 |
| Conditions | Atrial Fibrillation; Heart Murmurs |
| Intervention | Rapid Rhythm Handheld 8-lead ECG Device (device) |
| Lead sponsor | Manchester University NHS Foundation Trust |
| Sponsor type | OTHER_GOV |
| ClinicalTrials.gov | NCT03501836 |
2. Clinical Question
The central diagnostic question was whether ECG recordings obtained with the Rapid Rhythm handheld 8-lead ECG system could reproduce the findings of control ECGs, both when the control ECG was normal and when it contained abnormalities.
Population
Heart patients represented in the registry under the conditions atrial fibrillation and heart murmurs.
Intervention
Rapid Rhythm Handheld 8-lead ECG Device.
Comparator / reference
The primary endpoint definitions use a control ECG as the reference against which the Rapid Rhythm ECG is matched.
Primary question
What percentage of ECGs have Rapid Rhythm findings that match the corresponding normal or abnormal findings of the control ECG?
3. Trial Design
The registry classifies CHEST Extension as a single-group study with one arm, no masking, and a diagnostic primary purpose. This is fundamentally different from a randomized comparative treatment trial: there is no randomized treatment-versus-control contrast in the registered design.
Rapid Rhythm
- Rapid Rhythm Handheld 8-lead ECG Device
- Diagnostic primary purpose
- Single-group design
- No masking
Control ECG
- Used as the reference in the registered endpoint definitions
- Normal control ECGs are assessed for matching normal Rapid Rhythm ECGs
- Abnormal control ECGs are assessed for matching abnormalities on Rapid Rhythm ECGs
4. Trial Timeline and Status
Study start
The registry lists January 24, 2017 as the study start date.
Primary completion
The registry lists June 30, 2019 as the primary completion date.
Completed
The ClinicalTrials.gov record identifies the study status as COMPLETED.
5. Primary Endpoints
The registry specifies two primary endpoints, both assessed over an 18-month time frame. They are closely related but distinguish normal ECG matching from matching of abnormalities.
| Endpoint | Time frame | Registry definition |
|---|---|---|
| Capabilities of an 8 lead ECG system (Rapid Rhythm) - normal ECGs | 18 months | The percentage of ECGs where a normal control ECG is matched by a normal RR ECG. |
| Capabilities of an 8 lead ECG system (Rapid Rhythm) - abnormal ECGs | 18 months | The percentage of ECGs where abnormalities of the control ECG are matched by similar abnormalities of the RR ECG. |
Normal ECG endpoint
The endpoint asks for the percentage of ECGs in which a normal control ECG corresponds to a normal Rapid Rhythm ECG.
Abnormal ECG endpoint
The endpoint asks for the percentage of ECGs in which abnormalities on the control ECG are matched by similar abnormalities on the Rapid Rhythm ECG.
6. Statistical Methodology
The registry's primary endpoints are percentage-based diagnostic matching measures. Unlike a time-to-event endpoint, they do not require Kaplan-Meier estimation or a Cox proportional-hazards model. The natural statistical structure is based on counting how many ECG observations meet the specified matching criterion and expressing that count as a percentage of the relevant ECG observations.
Percentage matching
For either endpoint, the fundamental quantity can be represented as a proportion:
The numerator depends on the endpoint definition: a normal control ECG matched by a normal Rapid Rhythm ECG for the normal endpoint, or an abnormal control ECG matched by similar Rapid Rhythm abnormalities for the abnormal endpoint.
This formulation makes an important distinction between the number of observations that match and the percentage that match. The percentage is easier to compare across datasets of different sizes, but its precision still depends on how many ECG observations contribute to the denominator.
Binary classification framework
Each relevant ECG can conceptually be classified according to whether it satisfies the endpoint's matching definition. This creates a binary outcome for the purpose of calculating the percentage. For a normal control ECG, the relevant outcome is whether the Rapid Rhythm ECG is classified as normal. For an abnormal control ECG, the relevant outcome is whether the Rapid Rhythm ECG identifies a similar abnormality.
Confidence intervals for a percentage
When a diagnostic matching percentage is estimated from a finite number of observations, a confidence interval can be used to describe statistical precision. For a proportion p estimated from n observations, the interval is intended to quantify uncertainty arising from sampling rather than to describe the range of results that individual ECGs can produce.
Here, x represents the number of ECGs satisfying the matching criterion and n represents the number of relevant ECGs assessed. The registry reports the endpoint as a percentage rather than as a hazard ratio or time-to-event measure.
What this design does not require
Because the registry identifies one study arm and a diagnostic purpose, there is no registered randomized treatment contrast requiring a between-arm hypothesis test. There is also no registered time-to-event endpoint requiring survival analysis, no factorial structure, and no registered non-inferiority comparison in the information reported for this study.
7. Planned Analysis
No formal statistical analyses are posted to ClinicalTrials.gov for CHEST Extension. The registry does, however, define what the primary endpoints are intended to measure and the time frame over which they are assessed.
Normal ECG matching
The planned descriptive quantity is the percentage of ECGs for which a normal control ECG is matched by a normal Rapid Rhythm ECG over the 18-month endpoint time frame. A natural statistical presentation would pair the estimated percentage with its denominator and, where appropriate, a confidence interval for the underlying proportion.
Abnormal ECG matching
The second endpoint measures the percentage of ECGs for which abnormalities on the control ECG are matched by similar abnormalities on the Rapid Rhythm ECG over 18 months. The same basic proportion framework applies, but the relevant observations are those to which the abnormal-ECG endpoint definition applies.
8. Statistical Methods Explained
What does the normal-ECG percentage actually measure?
It measures the proportion of ECGs for which a normal control ECG is matched by a normal Rapid Rhythm ECG. A higher percentage would indicate more frequent matching under that definition, but the registry does not post a numerical result from which a specific performance level can be stated.
What does the abnormal-ECG percentage measure?
It measures the proportion of ECGs in which abnormalities identified on the control ECG are matched by similar abnormalities on the Rapid Rhythm ECG. The word similar is important because the endpoint is not simply defined as any abnormal reading; it is defined around matching the abnormalities of the control ECG.
Why is the denominator important?
A percentage without its denominator can be difficult to interpret statistically. For example, the precision of a percentage based on a small number of ECG observations is different from the precision of the same percentage based on many observations. The enrollment of 271 participants does not by itself establish the denominator for either primary endpoint because the registry describes the endpoints in terms of ECGs rather than explicitly stating that each participant contributes exactly one observation.
Why would a confidence interval be useful?
A confidence interval would communicate how precisely the observed matching percentage estimates the corresponding underlying proportion. A narrow interval would indicate greater statistical precision than a wide interval, all else equal. It would not mean that individual ECG classifications are uncertain over the interval.
Why is this not a hazard-ratio analysis?
Hazard ratios are used for time-to-event outcomes such as death, progression, or recurrence. The registered CHEST Extension endpoints are percentages of ECGs meeting diagnostic matching definitions over an 18-month time frame. Their statistical structure is therefore based on proportions rather than instantaneous event rates.
Why is the control ECG important?
The control ECG establishes the reference classification used by both primary endpoints. The statistical meaning of the matching percentage depends on this reference: a normal Rapid Rhythm ECG is considered a match in the first endpoint when the control ECG is normal, while a similar abnormality is considered a match in the second endpoint when the control ECG contains an abnormality.
Why should normal and abnormal matching be reported separately?
The two endpoints address different diagnostic situations. A device can perform differently when reproducing normal recordings than when identifying and reproducing abnormalities. Keeping the endpoints separate preserves the distinction defined in the registry rather than collapsing different diagnostic questions into one overall percentage.
9. Interpreting Diagnostic Percentages
A reported matching percentage would describe the proportion of relevant ECG observations that satisfied the corresponding registered matching definition. For example, a normal-ECG matching percentage would quantify how often the Rapid Rhythm reading was normal when the control ECG was normal.
The percentage would not automatically represent an overall probability that a patient has or does not have a cardiac condition. It would also not by itself establish sensitivity, specificity, predictive value, or diagnostic accuracy unless the study's complete classification framework supports those particular measures.
The point estimate alone does not communicate statistical uncertainty. A percentage calculated from many observations can be estimated more precisely than the same percentage calculated from few observations. The denominator and an appropriate confidence interval are therefore important components of a complete diagnostic performance analysis.
The registry specifies an 18-month time frame for both primary endpoints. That time frame identifies when the endpoint is assessed in the study record; it should not be confused with a claim that the diagnostic device has an 18-month survival effect or an 18-month event rate.
10. Analysis Population and Unit of Analysis
The registry reports an enrollment of 271 and identifies a single study arm. The primary endpoints, however, are expressed in terms of ECGs. This distinction matters statistically because the participant count and the number of ECG observations need not be identical.
| Statistical element | What the registry establishes |
|---|---|
| Participants enrolled | 271 |
| Study arms | 1 |
| Unit named in primary endpoint definitions | ECGs |
| Primary endpoint type | Percentage of ECGs satisfying a matching definition |
| Endpoint time frame | 18 months |
This distinction also affects how uncertainty should be calculated. If multiple ECG observations come from the same participant, those observations may not be statistically independent. A simple binomial confidence interval treats observations as independent unless the analysis accounts for within-participant correlation. The registry information presented here does not specify the correlation structure or a statistical adjustment for repeated observations.
11. Diagnostic Agreement Versus Clinical Accuracy
The wording of the primary endpoints centers on matching the control ECG. That is a specific form of diagnostic agreement. It should be distinguished from broader claims about clinical diagnostic accuracy.
| Concept | Meaning in this context |
|---|---|
| Matching | Whether the Rapid Rhythm ECG satisfies the registered definition of matching the corresponding control ECG. |
| Agreement | General agreement between two measurement or classification systems. |
| Sensitivity | Proportion of true positive cases identified by a test; requires a defined positive reference classification. |
| Specificity | Proportion of true negative cases correctly identified; requires a defined negative reference classification. |
| Positive predictive value | Proportion of positive test classifications that are truly positive; depends on the underlying prevalence and reference classification. |
| Negative predictive value | Proportion of negative test classifications that are truly negative; also depends on prevalence and reference classification. |
The registered endpoints support interpretation in terms of matching percentages. Additional diagnostic measures would require the corresponding classification counts and definitions. Those elements are not reported in the registry information available for this analysis.
12. Limitations
- No posted statistical analyses: the ClinicalTrials.gov record does not post formal statistical analyses or numerical results for the two primary endpoints.
- Single-group design: the study is not randomized and does not contain a registered treatment-versus-control allocation.
- Participant versus ECG counts: enrollment is reported as 271 participants, whereas the primary endpoints are defined using ECG observations. The registry information does not establish the number of ECGs contributing to each endpoint.
- Reference definition: the endpoints depend on the control ECG, but the available record does not provide a full diagnostic classification table from which sensitivity, specificity, predictive values, or other measures could be calculated.
- Repeated observations: if participants contributed more than one ECG, observations from the same participant could be correlated. The registry information does not specify whether such clustering was accounted for statistically.
- No reported confidence intervals: without posted endpoint results, the precision of the matching percentages cannot be evaluated from the registry record.
- No comparative treatment effect: the single-group design does not support interpretation in terms of a randomized treatment effect.
- Diagnostic endpoint scope: matching the control ECG is narrower than establishing the complete clinical diagnostic performance of a device across all possible patient populations and clinical settings.
13. Why This Trial Matters Statistically
CHEST Extension is a useful teaching example because its statistical structure is substantially different from the randomized survival analyses that dominate many clinical-trial reports. The primary questions are based on diagnostic matching percentages, making the definition of the reference measurement, the unit of analysis, and the denominator central to interpretation.
| Concept | How it appears in CHEST Extension |
|---|---|
| Single-group design | One registered study arm with 271 enrolled participants. |
| Diagnostic purpose | The registry identifies DIAGNOSTIC as the primary purpose. |
| Reference measurement | The primary endpoints compare Rapid Rhythm ECG findings with control ECG findings. |
| Binary classification | Each relevant ECG can conceptually be classified according to whether it satisfies the registered matching definition. |
| Percentage estimation | The primary endpoints are expressed as percentages of ECGs meeting their respective definitions. |
| Statistical precision | Confidence intervals for proportions can quantify uncertainty around a matching percentage when the underlying observation count is known. |
| Unit of analysis | The endpoints refer to ECGs rather than simply to the 271 enrolled participants. |
| Reference-standard interpretation | The meaning of a match depends on the control ECG classification. |
14. What a Complete Statistical Report Would Show
For a diagnostic matching study of this structure, a statistically informative results presentation would normally go beyond a single percentage. It would identify the number of relevant ECG observations, the number satisfying the matching criterion, the resulting percentage, and an appropriate measure of uncertainty.
Numerator
The number of ECG observations satisfying the registered matching definition.
Denominator
The number of relevant ECG observations against which the matching percentage is calculated.
Point estimate
The resulting proportion or percentage of ECGs meeting the endpoint definition.
Precision
An appropriate confidence interval showing statistical uncertainty around the matching percentage.
If repeated ECG observations were contributed by individual participants, a complete analysis could also need to consider within-participant correlation. The appropriate approach would depend on how ECG observations were collected and how the endpoint was operationalized.
15. Statistical Interpretation of the Two Primary Questions
Normal ECGs
The first endpoint asks whether a normal control ECG is matched by a normal Rapid Rhythm ECG. Statistically, this is a proportion of relevant normal-control observations satisfying a predefined binary matching criterion. The endpoint therefore focuses on agreement in the normal category rather than on a general treatment effect.
Abnormal ECGs
The second endpoint asks whether abnormalities on the control ECG are matched by similar abnormalities on the Rapid Rhythm ECG. This is another proportion, but its clinical meaning is different because the observations are defined by the presence of abnormalities on the control ECG.
16. Trial Design Features Relevant to Statistical Interpretation
| Design feature | Registry status | Statistical implication |
|---|---|---|
| Allocation | NA | No randomized allocation comparison is identified. |
| Design model | SINGLE_GROUP | The analysis centers on performance within one study group. |
| Masking | NONE | The registry does not identify a masking scheme. |
| Primary purpose | DIAGNOSTIC | The principal statistical questions concern diagnostic device performance. |
| Number of arms | 1 | There is no second randomized study arm for a comparative effect estimate. |
| Phase | NA | No clinical-development phase is specified in the registry record. |
17. What Is and Is Not Established by the Registry Record
| Question | Registry information available |
|---|---|
| How many participants were enrolled? | 271 |
| What device was evaluated? | Rapid Rhythm Handheld 8-lead ECG Device |
| How many study arms were registered? | 1 |
| What were the primary endpoints? | Normal ECG matching and abnormal ECG matching |
| What was the endpoint time frame? | 18 months |
| Were numerical primary endpoint results posted? | No formal statistical analyses are posted to ClinicalTrials.gov. |
| Were hazard ratios reported? | No. |
| Were confidence intervals reported? | No. |
| Were p-values reported? | No. |
18. Sources
- ClinicalTrials.gov: NCT03501836 — Wireless Handheld 8-lead ECG Device Performance in Heart Patients (Rapid Rhythm).
Continue through the Clinical Biostats knowledge graph
Explore the broader Clinical Biostats collection of clinical-trial analyses, statistical tutorials, and quantitative methods.
19. Record Summary
CHEST Extension is a completed single-group diagnostic study of the Rapid Rhythm handheld 8-lead ECG device, with 271 participants and two registered primary endpoints assessed over 18 months. Both endpoints are defined as percentages of ECGs that match findings from a control ECG: one endpoint addresses normal ECGs and the other addresses similar abnormalities. The statistical structure is therefore fundamentally one of diagnostic matching and proportion estimation rather than randomized treatment comparison or time-to-event analysis.