Image Quality Assessment for Screening and Diagnostic Mammography: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Image Quality Assessment for Screening and Diagnostic Mammography (NCT02786004), a non-randomized na study. The primary outcome is Number of Participants With Acceptable Overall Clinical Image Quality.

2. Clinical Question

3. Trial Design

NCT02786004, titled “Image Quality Assessment for Screening and Diagnostic Mammography,” is recorded as completed. The official title is “Mammography Image Quality Assessment Reading NextGen Images for Screening and Diagnostic Use (MAGNIFI).” The sponsor is GE Healthcare. The study start is recorded as 2016-02 and primary completion as 2016-05. Phase is listed as NA, which is a registry classification rather than a reported phase assignment.

The design is described as parallel, non-randomized, and unmasked. The listed arms are Full Field Digital Mammography and Digital Breast Tomosynthesis. Their descriptions characterize the imaging as 2-dimensional and 3-dimensional, respectively. The registry does not report how participants or images were assigned to arms, nor does it describe a randomized allocation procedure. Because allocation was non-randomized, the arms should not be treated as a like-for-like control comparison without further information about their formation and comparability.

Enrollment is recorded as 24. The primary outcome concerns participants with acceptable overall clinical image quality, assessed by MQSA qualified radiologists. The registry describes a timeframe at enrollment completion approximately 3 months post initiation. It does not report the number of readers, the reading procedure, whether the same readers assessed both imaging types, or how reader judgments were combined.

The record lists a secondary outcome concerning device-related adverse events, serious adverse events, unanticipated device effects, and device defects by overall occurrence. It gives no detailed protocol, analysis plan, or description of study conduct beyond these registry entries. The record is marked completed, but that status alone does not supply the missing outcome results.

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.

Overall Study

Reasons for not completing this period:

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.

6. Primary Endpoint

The registry lists the primary outcome as Number of Participants With Acceptable Overall Clinical Image Quality, measured over At enrollment completion approximately 3 months post initiation.

The primary endpoint is the number of participants with acceptable overall clinical image quality. It is a participant count: the underlying judgment is whether image quality is acceptable, and the endpoint expresses how many participants meet that criterion. The registry says that MQSA qualified radiologists assessed overall clinical image quality and image characteristics for FFDM and DBT. It cites the Guidance for Industry and FDA Staff - Class II Special Controls Guidance Document: Full Field Digital Mammography System (2012) as the reference for the assessment.

Image quality is relevant to a study of mammography devices because the study’s stated purpose is to determine the image quality of images made using a new investigational medical imaging device. A defined acceptability assessment can make the outcome more interpretable than an unspecified general impression. However, the registry record does not give the detailed acceptability criteria, reader instructions, adjudication procedure, or outcome values. Those omissions limit the ability to understand exactly how the judgment was applied and to assess the result.

The record gives a timeframe of at enrollment completion, approximately 3 months post initiation. It does not report a numerical count, a denominator for the primary endpoint, or an estimate comparing arms. Without those results, the endpoint description establishes what was intended to be measured, but not what the assessment found.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry record identifies the design as parallel and non-randomized, with no masking. It does not report a statistical analysis plan, a model used for the primary outcome, a test, an interval-estimation method, or any adjustment for participant or reader characteristics. Those details are important because they determine what comparison is estimated and how uncertainty is quantified. Their absence means that the analysis approach cannot be reconstructed from the registry record.

What does a participant count represent?

The primary measure is the number of participants with acceptable overall clinical image quality. Conceptually, each participant contributes to a category based on the acceptability judgment, and the outcome is the count meeting the criterion. A count is easiest to interpret alongside its denominator: the denominator indicates how many participants were assessed, and the count indicates how many met the criterion. The registry record does not report primary-outcome counts or denominators for the two arms. The adverse-event risk counts are not a substitute for primary-outcome denominators.

General statistical explanation: a binary outcome has two possible participant-level states, such as meeting or not meeting an acceptability criterion. Analysts may describe the count in each state, or estimate the probability of the outcome. A probability estimate is more comparable across groups of different sizes than a count alone, but its interpretation depends on a clear denominator and a defined population. Neither arm-specific primary counts nor proportions are reported in this registry record.

Why does non-randomized allocation matter?

Random allocation is a way to reduce systematic differences between groups at the point of assignment. When allocation is non-randomized, group differences may be associated with factors other than the imaging approach. General statistical explanation: this can complicate causal interpretation, because an observed contrast may reflect both the exposure of interest and group composition. Statistical adjustment can sometimes address measured differences, but it cannot automatically remove bias from unmeasured differences. The registry record does not describe baseline characteristics, an adjustment strategy, or an analysis intended to address such differences. It also explicitly notes that the arms are not a like-for-like control comparison.

The absence of masking can matter when outcomes involve judgment. Here, readers assessed image quality and image characteristics, but the registry lists masking as none. The record does not say whether readers knew the imaging type when assessing images, whether image presentation was standardized, or whether judgments were reviewed or adjudicated. General statistical explanation: a reader’s awareness of the assessed condition can potentially influence subjective assessment. The available description does not establish whether that occurred or how it was handled.

How should uncertainty be understood?

General statistical explanation: a confidence interval presents an estimate together with an indication of its uncertainty under the chosen statistical method. It helps readers distinguish a precise estimate from one compatible with a broad range of possible values. Its width depends on features such as the amount of information and variability in the data, as well as the method used. A p-value addresses compatibility between data and a specified null model; it is not the probability that a treatment works, nor does it measure the size or importance of an effect. No primary estimate, confidence interval, or p-value is reported in the registry record, so these aspects of precision and statistical evidence cannot be assessed for the primary endpoint.

What do the planning assumptions mean?

The record describes a sample-size planning illustration, not the trial’s own design justification. Its stated assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, a normal approximation, and an effect to detect expressed as a stated fraction of the observed effect. The illustration is described as computed from the observed effect. These assumptions do not establish that the original study was designed or powered using that approach. The registry does not report a protocol sample-size justification or the specific statistical planning choices used for the study itself.

General statistical explanation: a planning calculation asks how much information might be needed to detect an effect under chosen assumptions. Changing the effect targeted or the assumptions changes the planning result. Such a calculation is not observed power and is not evidence that an observed result is true or precise. Because the registry record does not include primary outcome results or a numerical observed effect, the planning assumptions alone cannot be used to infer the study’s actual result.

Can the analysis be reproduced?

Reproduction would require, at minimum, the outcome counts and denominators, details of the analysis population, the statistical method, and relevant handling of missing or indeterminate assessments. For this study, those details are not reported in the registry record. The readers’ acceptability judgments and the way they were converted into participant-level outcomes are also not described in sufficient detail to reconstruct the endpoint. The registry therefore permits a description of the intended measurement and broad design, but not an independent recalculation of the primary comparison.

10. Exact Estimates for Each Arm

No two-arm comparison is shown for this trial, but each arm's result can still be summarized with exact interval estimates computed from the posted counts. The Clopper-Pearson interval is exact and conservative; the Wilson interval is a common closer-to-nominal alternative; the Bayesian credible interval uses a uniform Beta(1, 1) prior. For small groups the three can differ noticeably.

R code and output

arm    <- c("Full Field Digital Mammography", "Digital Breast Tomosynthesis")
n      <- c(14, 9)
events <- c(14, 9)
for (i in seq_along(n)) {
  ci <- binom.test(events[i], n[i])$conf.int          # Clopper-Pearson exact interval
  cr <- qbeta(c(0.025, 0.975), 1 + events[i], 1 + n[i] - events[i])   # Bayesian, uniform Beta(1,1) prior
  wi <- (function(x, n) { p <- x / n; z <- qnorm(0.975); d <- 1 + z^2 / n
    c((p + z^2 / (2 * n) - z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d, (p + z^2 / (2 * n) + z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d) })(events[i], n[i])
  cat(sprintf("ARM|%s|%d|%d|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f\n", arm[i], n[i], events[i], 100 * events[i] / n[i],
              100 * ci[1], 100 * ci[2], 100 * wi[1], 100 * wi[2], 100 * cr[1], 100 * cr[2]))
}

ARM  Full Field Digital Mammography  14  14  100.0  76.8  100.0  78.5  100.0  78.2  99.8
ARM  Digital Breast Tomosynthesis  9  9  100.0  66.4  100.0  70.1  100.0  69.2  99.7

11. Multiplicity, Missing Data and Analysis Populations

The registry lists 2 outcome measures and identifies the participant count with acceptable overall clinical image quality as the primary outcome. It also names a secondary outcome concerning device-related adverse events, serious adverse events, unanticipated device effects, and device defects by overall occurrence. The record does not report whether the study prespecified additional comparisons, how many comparisons were planned, or whether any multiplicity adjustment was used.

General statistical explanation: when many outcomes or comparisons are tested, the chance of at least one apparently unusual result can increase. A prespecified analysis plan and suitable interpretation of multiple results can help readers understand which findings were primary and how the collection of tests was handled. The registry record does not give enough analysis detail to determine whether or how this issue was addressed. It does not report interim analyses or stopping rules.

Missing-data methods are not reported in the registry record. The record does not state whether any primary assessments were missing, how missing or indeterminate reader judgments were handled, or whether participants with incomplete data were included in an analysis. Analysis-population definitions are also not reported. The enrollment figure does not establish how many participants contributed to each endpoint or analysis.

These omissions affect interpretation because the endpoint result depends on which participants and assessments are included, and because missingness can influence estimates if it is related to image quality or other factors. General statistical explanation: complete-case analysis, imputation, and other approaches make different assumptions about missing information. Without a reported approach and the relevant counts, readers cannot judge those assumptions or determine whether missing data may have altered the findings.

12. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

The registry safety record lists Full Field Digital Mammography with 14 participants at risk and Digital Breast Tomosynthesis with 10 participants at risk. For Full Field Digital Mammography, the listed counts are 0 serious adverse events, 0 other adverse events, and 0 deaths. For Digital Breast Tomosynthesis, the listed counts are also 0 serious adverse events, 0 other adverse events, and 0 deaths. The record lists no top serious-event terms.

These are reported counts in the registry record. They do not describe event rates or provide a comparative safety estimate, and the record does not give further details about duration or completeness of safety follow-up, event definitions, or methods of collection. The risk counts belong to the safety entry and should not be treated as primary image-quality denominators.

General statistical explanation: an observed count of zero means that no events of the specified type were recorded in the listed group for the reported record. It does not establish that the underlying event risk is zero. The precision of any safety estimate depends on the amount and quality of follow-up and the number of observations, as well as on how events are defined and collected. Those details are not reported in the registry record, so the zero counts should be interpreted narrowly as recorded counts, not as proof of absence of risk.

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

The registry record establishes that this completed study was designed to assess mammography image quality for Full Field Digital Mammography and Digital Breast Tomosynthesis. It identifies a primary outcome based on radiologists’ judgments of acceptable overall clinical image quality and describes the imaging types as 2-dimensional and 3-dimensional. It also records enrollment of 24 and reports safety counts for the two listed groups.

The record does not report the numerical primary endpoint results, a comparison estimate, a confidence interval, or a p-value. It therefore does not show whether image quality was acceptable for more participants in either arm, how large any difference might have been, or how precisely a difference was estimated. The absence of these statistics is not evidence of no difference. It is a limit on what can be concluded from the registry entry.

The non-randomized allocation and lack of masking also shape interpretation. Even if a comparison were available, the record does not describe enough about group formation, reader procedures, participant characteristics, or statistical adjustment to determine how those features might affect the result. In general, a descriptive difference between non-randomized groups does not automatically establish that the imaging approach caused that difference.

The safety counts report no listed serious adverse events, other adverse events, or deaths in either group. These entries do not establish a zero underlying risk, and there is no further safety follow-up information in the registry record to support a broader claim.

Overall, the available record supports a description of the study question, design, endpoint definition, and recorded safety counts. It does not support a conclusion about comparative image quality or a broad claim about device performance. Those conclusions would require the missing outcome results and sufficient analysis and measurement details to assess the comparison and its uncertainty.

14. Why This Trial Matters Statistically

This study raises a useful statistical distinction between defining an outcome and reporting its result. The registry clearly identifies a participant-level image-quality question and names the readers who assessed it. Yet the numerical result for that primary outcome is not reported in the registry record. Without counts and denominators, readers cannot describe the observed acceptability outcome or assess its precision. A well-defined question is necessary for interpretation, but it cannot replace the result itself.

The study also illustrates why design labels matter. The registry describes a parallel, non-randomized comparison and explicitly warns that the arms are not a like-for-like control comparison. That qualification affects what a contrast can mean: group differences may influence the observed comparison, and the registry does not describe adjustment or other safeguards against such differences. This is a question of interpretation, not merely a technical detail.

Measurement also deserves attention when an endpoint depends on reader judgment. The record identifies MQSA qualified radiologists and cites a guidance document, but does not describe reader masking, rating procedures, or how assessments were combined. Those details can help establish how consistently and reproducibly the outcome was measured. Their absence leaves uncertainty about the operational details behind the endpoint.

Finally, the record separates planning illustrations from trial design. The listed assumptions describe a planning calculation based on an observed effect, not a reported protocol justification or evidence about the outcome. This distinction helps prevent planning quantities from being mistaken for observed evidence or observed power.

These features make the registry record valuable for understanding the intended question and important for recognizing what remains unknown. The missing primary result and analysis details limit statistical interpretation, while the stated design cautions against treating the arms as equivalent controls.

15. 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.

The most consequential limitation of the registry entry is that it does not report numerical results for the primary image-quality outcome. Counts, denominators, effect estimates, and uncertainty measures are not available for that endpoint. The registry also lists 0 analyses, so no primary statistical analysis is described there. These omissions prevent readers from evaluating the size or precision of the main comparison.

The design is non-randomized, and the record warns that the arms are not a like-for-like control comparison. It does not describe how participants or images entered each arm, the groups’ relevant characteristics, or any adjustment intended to address differences. The record also lists no masking. Since the outcome involves radiologist judgment, the absence of detail about reader awareness, assessment procedures, and review of disagreements limits understanding of measurement.

The endpoint description cites a guidance document but does not give the full operational definition of acceptability, reader instructions, or the method for combining assessments into a participant-level count. The registry does not report the primary analysis population, missing-data handling, or whether any assessments were incomplete or indeterminate. These details are needed to interpret a count and understand which observations it represents.

The safety record gives zero counts for specified event categories, but it does not provide further information on follow-up, event ascertainment, or the completeness of reporting. The record also does not report interim analyses, stopping rules, or the study’s own sample-size justification. Its planning assumptions are not a substitute for those protocol details.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Image Quality Assessment for Screening and Diagnostic Mammography — NCT02786004.
• Registry record last updated 2018-11-07; status last verified 2018-10.
• Results first posted 2017-07-21.
• 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.

19. Record Summary

NCT02786004, “Image Quality Assessment for Screening and Diagnostic Mammography,” is recorded as a completed, non-randomized, unmasked parallel study sponsored by GE Healthcare. The record lists enrollment of 24, a start of 2016-02, and primary completion in 2016-05. The arms are Full Field Digital Mammography, described as 2-dimensional breast imaging, and Digital Breast Tomosynthesis, described as 3-dimensional breast imaging.

The primary outcome is the number of participants with acceptable overall clinical image quality, assessed by MQSA qualified radiologists. Numerical primary results and analyses are not reported in the registry record. The safety entry lists 0 serious adverse events, 0 other adverse events, and 0 deaths in each arm, with 14 and 10 participants at risk, respectively. These counts do not establish a zero underlying risk. The record cautions that the non-randomized arms are not a like-for-like control comparison.