Reader Study to Demonstrate That Use of ClearRead Confirm is Superior to the Use Standard AP/PA X-ray Image: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Reader Study to Demonstrate That Use of ClearRead Confirm is Superior to the Use Standard AP/PA X-ray Image (NCT01655329), a registered phase not stated study. The primary outcome is Time for Completion of Tasks.

2. Clinical Question

The trial's stated primary interest was interpretation time. The registered endpoint was time for completion of tasks, measured in seconds over an 8 hours timeframe. Its description specifies that outlying times were excluded when calculating means and standard errors, with outliers defined as values more than three standard deviations from the mean time.

The research question concerns the performance of image interpretation under the two image presentations, not a clinical outcome for patients. The registry record does not report results for either the primary time endpoint or the stated accuracy question. It therefore does not establish whether interpretation time differed between image groups, how large any difference was, or how precise an estimate might be. The allocation method and the details of how the two image groups were compared are also not reported in the registry record.

3. Trial Design

This completed reader study was sponsored by Georgetown University and registered as NCT01655329. The registry record lists an enrollment of 10 and radiologists as the participants in the adverse-event group. The study began in 2012-08, with primary completion in 2012-12. Its status is COMPLETED.

The study concerned interpretation of chest radiographs containing tubes, lines, and wires. The registry describes a standard chest radiograph group and a modified chest radiograph group. The modified images were intended to increase visibility of those devices using ClearRead Confirm. Radiologists were asked to identify device tips, and the primary endpoint was the time required to complete those tasks. The study summary also identifies accuracy of detection as an interest.

The registry record says each radiologist interpreted only half (158) chest radiographs and that all radiographs were interpreted by a combination of the 10 radiologist participants. It does not provide further detail about which readers interpreted which images, whether a reader viewed both standard and modified versions of any image, or how presentation order was handled. The allocation method is not stated. The record cautions that the arms are not a like-for-like control comparison, so the labels should not be read as evidence of randomized allocation or balanced groups.

The primary endpoint is described as a mean with standard error. The registry specifies exclusion of outlying values, defined as times more than three standard deviations from the mean time, from the mean and standard-error calculations. The record lists three outcome measures but no analysis entries. It does not report a statistical analysis plan, a sample-size justification, interim analyses, or the methods used to compare the image groups. Those design details are not reported in the registry record.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Before assignment: 54 cases were used to train the radiologists on the use of the data recording computer and to make them familiar with the appearance of the processed chest radiographs. These cases were not used in the test of reader performance.

Overall Study

Analysis population for the primary outcome, as defined in the registry: Each radiologist interpreted only half (158) chest radiographs. All of the radiographs were interpreted by a combination of the 10 radiologist participants.

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

Population: Board Certified Radiologists

6. Primary Endpoint

The registry lists the primary outcome as Time for Completion of Tasks, measured over 8 hours.

The primary endpoint, “Time for Completion of Tasks,” measures the time required to identify the tips of tubes, lines, and wires on chest radiographs. It is recorded in seconds, and the registry describes the outcome as a mean with standard error over an 8 hours timeframe. Interpretation time was selected because the study summary identifies time used to interpret the radiograph as its primary interest. The study also aimed to determine whether detection accuracy changed, but accuracy is not the registered primary endpoint described in this record.

Time is a continuous measure: each recorded interpretation can take a range of values rather than falling into a small set of categories. A group mean gives an average time, while a standard error describes the statistical uncertainty around that estimated mean. General explanation: a mean can make a collection of times easier to summarize, but it can be affected by unusually long or short observations. The registry addresses this by excluding values more than three standard deviations from the mean time when calculating the mean and standard error.

That exclusion rule matters to interpretation. The posted mean, if reported, would describe the retained times under the registry's rule, not necessarily every recorded interpretation time. The record does not give the number of excluded observations, the mean times, standard errors, or a comparison estimate. Thus, the endpoint definition is available, but the observed result and its precision are not reported in the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry gives limited information about the statistical methods. It defines the primary outcome as a mean of a continuous time measure and specifies standard error as the dispersion measure. It also defines an outlier exclusion rule. However, it does not report a comparison method, model, confidence interval, hypothesis test, or analysis result. The explanations below describe general statistical concepts relevant to interpreting this type of outcome; they are not claims about methods used in the trial.

What does a mean describe?

A mean is an average of observed values. For this study, the values are the times taken to complete the device-tip identification tasks. A mean can provide a compact description of typical time, but it does not show the full distribution. For example, different sets of times can have the same mean while differing in spread or in the presence of unusually long observations. The registry's outlier rule therefore affects what the calculated mean represents: it is the mean after excluding times more than three standard deviations from the mean time, according to the endpoint description.

What does the standard error describe?

Standard error describes the uncertainty in an estimated quantity, such as a mean, due to sampling variation. It is not the same as the spread of individual interpretation times. A standard error is useful for understanding how precisely a mean has been estimated, but it cannot be interpreted fully without the estimate and the analysis context. The registry names standard error as the dispersion measure, but does not post its value for the primary endpoint. It also does not give a confidence interval.

General explanation: a confidence interval combines an estimate with an uncertainty range to show values compatible with the data under the method's assumptions. Its width reflects precision; a wider interval indicates greater uncertainty than a narrower interval. A confidence interval does not by itself establish causation or eliminate bias in how groups were formed. No confidence interval for interpretation time is reported in the registry record.

Why does the comparison structure matter?

In a reader study, multiple readings may be associated with the same radiologist, the same radiograph, or both. As a general statistical matter, readings that share a reader or image can be correlated. Correlation means that the observations may contain overlapping information; treating them as fully independent can make uncertainty appear smaller than it is. Methods for clustered or repeated observations can account for such dependence when the study design and data structure are known. The registry describes the combination of radiologists and radiographs but does not report the statistical method or enough allocation detail to determine how dependence was handled.

There is also a distinction between comparing two sets of readings from the same reader and comparing readings from different readers. Those structures call for different ways of estimating uncertainty. The registry does not say whether the readers assessed both types of image, whether images were paired across conditions, or how the allocation was conducted. Since allocation is not stated and the record cautions that the arms are not a like-for-like control comparison, a reader should not assume a paired or randomized analysis.

What is the role of the outlier rule?

The endpoint description says values more than three standard deviations from the mean time were defined as outliers and excluded from the mean and standard-error calculations. This is a prespecified-sounding definition in the registry description, but the record does not establish when or how the rule was applied, how many values it affected, or whether additional analyses examined its influence. In general, excluding extreme values can make a mean less sensitive to a few unusually high or low measurements, but can also change the result and reduce the amount of data represented. The result should therefore be understood in light of the rule, with the number excluded and sensitivity of conclusions being useful supporting information. These details are not reported in the registry record.

What can planning calculations tell a reader?

General explanation: sample-size planning uses assumptions about the effect a future study aims to detect, the variability of the outcome, the significance threshold, allocation, and the statistical method. A planning illustration is not a result from the trial and is not observed power. The stated assumptions for the illustrations are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation; the effect to detect is a stated fraction of the observed effect. The illustrations are planning calculations, not evidence that the completed study detected an effect. The registry record does not give an observed effect value or the resulting participant requirements in the registry record for this analysis.

Finally, the registry lists no analysis entries. It does not state whether any unadjusted or adjusted analyses were performed, whether results were examined at interim points, or how missing readings were treated. These omissions prevent a reader from reproducing or evaluating the statistical analysis. They should be distinguished from the endpoint definition, which does specify the outcome scale and the stated outlier rule.

10. Confidence Intervals for Each Arm

No two-arm comparison is shown for this trial, but each arm's mean can be given with a t-based 95% confidence interval computed from the posted mean, spread and number of participants.

R code and output

arm <- c("Standard Chest Radiographs", "Modified Chest Radiographs")
n   <- c(10, 10)
m   <- c(27.68, 22.9)
sp  <- c(3.32, 3.32)
sd  <- sp * sqrt(n)      # standard errors converted to SDs
for (i in seq_along(n)) {
  se <- sd[i] / sqrt(n[i]); h <- qt(0.975, n[i] - 1) * se
  cat(sprintf("CARM|%s|%d|%.3f|%.3f|%.3f|%.3f\n", arm[i], n[i], m[i], sd[i], m[i] - h, m[i] + h))
}

CARM  Standard Chest Radiographs  10  27.680  10.499  20.170  35.190
CARM  Modified Chest Radiographs  10  22.900  10.499  15.390  30.410

11. Multiplicity, Missing Data and Analysis Populations

The registry record lists three outcome measures, but it does not give a statistical analysis entry. The primary endpoint is time for completion of tasks; the study summary additionally says that detection accuracy was of interest. The record does not explain how the three outcomes were prioritized for statistical testing, whether any outcomes were secondary, or whether a multiplicity adjustment was planned. Multiplicity is a general concern when several outcomes or comparisons are tested: the more opportunities there are to identify a chance finding, the more carefully the analysis plan and interpretation need to address that issue. No adjustment or testing strategy is reported, so it is not possible to assess how multiple outcomes were handled.

The registry does not describe missing-data handling or define an analysis population for the primary endpoint. It does not report the number of completed readings included in the time analysis, the amount of missing data, reasons for missing readings, or whether any observations were excluded for reasons other than the stated outlier rule. The endpoint description specifies exclusion of times more than three standard deviations from the mean time from mean and standard-error calculations, but does not report how many times met that definition.

These omissions limit interpretation of both the estimate and its uncertainty. Missingness can matter if the availability of a reading is related to its time or to the image condition. Likewise, outlier exclusions can change the set of observations contributing to the reported mean. The record does not provide enough information to judge either issue or to reproduce the primary analysis. The fact that the trial status is COMPLETED does not resolve these reporting gaps.

12. What the Result Does, and Does Not, Mean

The registry record establishes the study's question and endpoint definition: radiologists interpreted standard and ClearRead Confirm-modified chest radiographs, and the primary measure was time to identify device tips. It also describes an outlier rule for calculating mean times and standard errors. These details define what the study sought to measure, but do not provide the observed result.

The record does not report the mean interpretation times, a between-group estimate, a confidence interval, a test result, or the detection-accuracy results mentioned in the study summary. It therefore cannot show the direction or magnitude of any difference, or how precisely a difference was estimated. The missing results should not be interpreted as evidence that the image presentations produced equal outcomes, nor as evidence that either presentation was superior.

The design information also limits what can be inferred. Allocation is not stated, and the record explicitly warns that the arms are not a like-for-like control comparison. The reader and image assignment details and the statistical handling of repeated readings are not reported. Even if an effect estimate were available, those details would be important in judging how directly the comparison reflects the image modification rather than other differences in study conditions.

General interpretation: an endpoint about reading time addresses speed, not necessarily accuracy, patient outcomes, or the overall quality of interpretation. The registry summary notes accuracy as an additional interest, but no accuracy results appear in the record. Thus, the stated study question is narrower than a claim about clinical benefit. The posted information supports description of the study design and measurement, but not a conclusion about the effect of ClearRead Confirm.

13. Why This Trial Matters Statistically

This study raises a useful statistical issue: an image-reading intervention can be evaluated using both interpretation time and accuracy, and those outcomes do not answer the same question. Time measures how long the task takes, while accuracy concerns whether device tips are detected correctly. A faster reading time would not, by itself, describe accuracy; an accuracy result would not by itself describe speed. The registry summary recognizes both interests, while identifying time as the primary endpoint.

The reader-study structure also makes the unit of analysis important. Multiple interpretations can be connected to the same radiologist or radiograph. As a general principle, statistical methods should reflect such dependence when estimating uncertainty. The record describes a combination of 10 radiologist participants interpreting the radiographs, but does not report the method used to account for reader or image effects. This is a reason to look for analysis details, not a basis to assume that the analysis was incorrect.

The outlier rule is another consequential detail. The registry defines which unusually distant times were excluded from the mean and standard-error calculations. Reporting such a rule helps readers understand how the endpoint was formed, but interpretation is stronger when the number excluded and the effect of the exclusion are also available. Those details are not reported in the registry record.

Finally, a completed study record can still lack posted results. Here, the endpoint and broad image groups are described, while the outcome estimates and analysis are not reported. That distinction matters statistically: a defined research question is not the same as an estimated effect, and planning calculations are not a substitute for observed results. Clear reporting of estimates, uncertainty, allocation, and analysis methods is needed to assess what the study found.

14. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

A central limitation of the available record is the absence of primary outcome results. There are no reported mean times, standard errors, comparative estimates, confidence intervals, or test results. The study summary's accuracy question also has no reported result. This prevents assessment of effect direction, size, precision, or consistency.

The registry cautions that the arms are not a like-for-like control comparison, and the allocation method is not stated. Further details about reader assignment, image assignment, presentation order, and whether the same radiologists assessed both image groups are not reported in the registry record. These gaps make it difficult to assess comparability or determine which statistical comparison would be appropriate.

The reader and image structure may create related observations, but the analysis method is not given. The record also does not describe missing readings, the analysis population, or how many times were excluded under the stated outlier definition. Because the primary endpoint is a mean after outlier exclusion, the effect of that rule cannot be evaluated without the excluded count or additional analysis information.

The record lists three outcome measures but no analysis entries. It does not report how multiple outcomes were addressed, whether analyses were adjusted, or whether interim analyses occurred. The protocol, sample-size justification, and statistical plan are not reported in the registry record. The enrollment and completion status do not fill in these missing analytic details.

These limitations concern what can be learned from the registry record; they should not be taken as evidence that the study itself did not collect the relevant information. The record does not establish whether additional results or methods are available elsewhere.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Reader Study to Demonstrate That Use of ClearRead Confirm is Superior to the Use Standard AP/PA X-ray Image — NCT01655329.
• Registry record last updated 2014-07-10; status last verified 2014-06.
• Results first posted 2014-07-10.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

18. Record Summary

NCT01655329 is a completed Georgetown University reader study of ClearRead Confirm-modified chest radiographs and standard chest radiographs. It enrolled 10 radiologist participants. The study began in 2012-08 and reached primary completion in 2012-12.

The primary endpoint was time for completion of tasks to identify the tips of tubes, lines, and wires, measured in seconds over an 8 hours timeframe. The registry describes the endpoint as a mean with standard error and says times more than three standard deviations from the mean were excluded from those calculations. The study summary also identifies detection accuracy as an interest.

The registry record does not state the allocation method, and it warns that the arms are not a like-for-like control comparison. It lists three outcome measures but no analysis entries. Primary outcome estimates, accuracy results, and statistical methods are not reported in the registry record.