Adult Accuracy Study of the Enlite 3 Glucose Sensor: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Adult Accuracy Study of the Enlite 3 Glucose Sensor (NCT02246582), a randomized na study. The primary outcome is Enlite 3 Sensor Accuracy Mean Absolute Relative Difference (MARD).

2. Clinical Question

The central question concerns measurement agreement: how far sensor readings were from the reference values, relative to those reference values, on average. The outcome description says results from the testing days were pooled for reporting. Consequently, the stated primary question is about the pooled accuracy measure across the specified testing days, rather than a separate reported estimate for each day.

The record also lists retrospective analyses involving an additional calibration and calibration conditions for the GSR. Those are secondary outcomes, not substitutes for a reported primary result. No numerical results for them are reported in the registry record. The question is therefore clearly defined at the level of the measurement target and reference method, but the posted results do not show the size or precision of the observed accuracy estimate.

3. Trial Design

This was a randomized, parallel-group study with no masking. The registry record lists enrollment of 118 and describes the study as completed. It lists type 1 diabetes and type 2 diabetes as conditions, and the study summary describes participants aged 14–75 years. The recorded study start was 2014-09 and primary completion was 2015-08.

Although participants were randomized to two groups, the primary outcome description says the data were collected as a whole and there was no intention to analyze the groups separately. The groups differed in the timing of frequent sample testing: Group A underwent FST at 30 mins, 50 hrs and 146 hrs, while Group B underwent FST at 14 hrs, 62 hrs and 158 hrs. The description identifies Enlite 3 sensors connected to GST3C, GST4C Transmitter, and GSR. The registry record does not explain the rationale for the different testing schedules.

The evaluation used Enlite 3 with two real-time devices named in the primary outcome description: the 640G Pump and Guardian Mobile. The stated sensor-use period was 168 hours (7 days). The primary measure pooled results from frequent sample testing on Days 1, 3 and 7 and compared sensor values with YSI plasma glucose values.

Randomization is a way to assign participants to groups by chance, but it does not establish that the study was designed to estimate a difference between those groups. In this record, the stated analysis approach is pooled data, not a separate group comparison. The record does not report a protocol, the allocation procedure in further detail, a sample-size justification, or a statistical analysis plan. Those design details should not be inferred from the randomized label alone.

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: 118 subjects consented and enrolled in the overall study. Of the 118, 89 subjects utilized the latest sensor version.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Even though participants were randomized to two groups, data was collected as a whole and there was no intention to analyze the two groups seperately.

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: Even though participants were randomized to two groups, data was collected as a whole and there was no intention to analyze the two groups seperately.

6. Primary Endpoint

The registry lists the primary outcome as Enlite 3 Sensor Accuracy Mean Absolute Relative Difference (MARD), measured over 7 Days.

The primary endpoint is Enlite 3 Sensor Accuracy Mean Absolute Relative Difference (MARD), reported as a mean with standard deviation over 7 Days. MARD summarizes the average size of the absolute difference between sensor glucose and YSI plasma glucose, expressed relative to the YSI reference value. The registry record gives the underlying calculation as the absolute difference divided by the YSI value, multiplied by 100, then averaged.

Absolute differences treat readings above and below the reference as deviations in magnitude, rather than allowing positive and negative differences to cancel. Scaling the difference relative to the reference makes the measure interpretable as a percentage of the reference value. In general, a smaller MARD represents a smaller average relative discrepancy from the reference; it does not, by itself, describe every individual reading or the full distribution of errors.

The outcome description specifies two real-time devices, the 640G Pump and Guardian Mobile, with minimum calibration requirements described as every 12 hours after the second calibration. The reference measurements were YSI plasma glucose values, and testing was scheduled on Days 1, 3 and 7. Results from those testing days were pooled for reporting. The registry record does not report the MARD estimate or its standard deviation, so the endpoint’s observed magnitude and variability cannot be assessed from the posted results.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does MARD describe?

MARD is a summary of the absolute relative difference between sensor values and a reference measurement. The registry record defines the reference as YSI plasma glucose and describes calculating the absolute difference, dividing by the YSI value, multiplying by 100, and taking the mean. Absolute values prevent readings on opposite sides of the reference from canceling each other in the average. Expressing differences relative to the reference gives the measure a percentage scale.

As a general statistical interpretation, MARD describes average discrepancy, not exact agreement for every reading. An average can be influenced by both the size and distribution of the observed discrepancies. Two sets of readings can have the same mean discrepancy while differing in how tightly or unevenly their individual errors are distributed. The standard deviation listed for the endpoint is intended to describe spread, but its value is not reported in the registry record.

Why use a reference measurement?

A measurement-accuracy evaluation needs a comparator that acts as the reference for judging the sensor readings. The record identifies YSI plasma glucose values as the gold standard for this evaluation. Each sensor value is therefore interpreted in relation to its corresponding reference value, rather than as a clinical outcome or as a randomized treatment assignment. The record does not provide details about the reference measurement process, pairing rules, or quality-control procedures, so those details are not reported in the registry record.

In general, a reference comparison can quantify the average discrepancy from the chosen reference, but it cannot establish properties that were not measured. MARD alone does not give the direction of discrepancies, the frequency of large discrepancies, or how the readings compare at particular glucose levels. Additional summaries or analyses would be needed to answer those questions. The registry record does not report such numerical summaries.

What does pooling across testing days imply?

The primary outcome description says testing took place on Days 1, 3 and 7 and that results from multiple testing days were pooled for reporting. Pooling produces a summary across the included testing observations rather than a separate reported estimate for each named day. In general, a pooled summary can conceal variation across time points: an overall mean does not reveal whether accuracy differed by day. Day-specific estimates and uncertainty would be needed to assess that pattern, and they are not reported in the registry record.

The groups had different FST schedules, but the primary outcome description says the data were collected as a whole and there was no intention to analyze the groups separately. Thus, the randomized parallel-group structure should not be read as evidence that the primary outcome was analyzed as a randomized comparison. The record does not describe a separate group effect estimate.

What do mean and standard deviation convey?

The endpoint is specified as a mean with standard deviation. In general, the mean is a central summary of the measured values, while the standard deviation describes how spread out those values are around the mean. Neither alone gives the precision of the mean estimate. A confidence interval can communicate uncertainty about an estimated mean, under the assumptions of the analysis. The registry record does not report a confidence interval, standard error, or numerical mean and standard deviation for the primary endpoint.

Repeated measurements can create statistical dependence: observations from the same participant may be more alike than observations from different participants. As a general principle, an analysis of repeated data should account for the structure relevant to its target estimate. The record does not report the number of readings per participant, the handling of repeated observations, or a model for the pooled measure. It is therefore not possible to determine from the posted information how within-participant dependence was addressed.

What can planning calculations say?

The registry record describes sample-size illustrations based on the observed effect, with a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. It also says that the effect to detect is expressed as a stated fraction of the observed effect. These are planning assumptions, not evidence about what the trial observed. The record does not report numerical illustration results or the observed effect value. Such illustrations should not be interpreted as observed power or as an analysis of the trial findings.

In general, planning calculations explore the size of a future study under specified assumptions about an effect, variability, significance level, and allocation. Their output is conditional on those assumptions. If the effect or variability used as a basis is uncertain, the resulting planning values are uncertain as well. Since the numerical values and the observed endpoint result are not reported in the registry record, no quantitative sample-size conclusion can be drawn from those assumptions alone.

10. Multiplicity, Missing Data and Analysis Populations

The registry record lists one primary outcome and two secondary outcomes. The secondary outcomes concern retrospective re-analysis of MARD with an additional calibration and retrospective analysis of MARD for the GSR with minimum and an additional calibration. Numerical results for these outcomes are not reported in the registry record.

When several outcomes or analyses are examined, the chance of obtaining at least one apparently notable result can increase if each is assessed independently. Statistical procedures for multiplicity can address this issue, depending on the analysis goals and prespecified plan. The record does not report a multiplicity adjustment, a testing hierarchy, or a statistical analysis plan. It is therefore not possible to determine how multiple outcomes or analyses were handled.

The outcome description says data from testing days were pooled, and the primary outcome description says the randomized groups were not intended to be analyzed separately. No day-specific estimates or group-specific estimates are reported. The record does not state whether additional exploratory analyses were conducted or how any such analyses were distinguished from planned analyses.

Missing-data handling and analysis populations are also not reported in the registry record. Although enrollment is listed as 118, the record does not say how many participants contributed primary outcome measurements, whether any measurements were excluded, or how missing or unusable readings were handled. It does not identify an analysis population such as all randomized participants or participants with evaluable measurements. These omissions limit the ability to assess how representative the reported analysis would be of enrollment, even if a numerical result were available.

The record reports no analyses and no numerical primary or secondary outcome findings. It does not report imputation, sensitivity analyses, exclusions, or a method for accounting for repeated measurements. These details should not be inferred from the endpoint description or the randomized design.

11. Safety Results

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

Other (non-serious) adverse events

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

Most frequent non-serious adverse event terms

The safety entry lists one group with 89 participants at risk for serious adverse events and zero affected by a serious adverse event. It also lists 89 at risk for other adverse events and five affected. The record does not list any serious adverse event terms in its serious-event term summary.

These are counts for the listed safety group, not a comparison between the randomized groups. The registry record does not give a safety follow-up period, event-level descriptions, or the relationship of the safety group to the primary analysis population. Enrollment is listed as 118, but the record does not explain how that enrollment relates to the 89 participants at risk in the safety entry. That difference cannot be resolved from the posted information.

The death value is not reported in the registry record. A zero count for participants affected by serious adverse events should not be treated as a report of zero deaths, because the death field is not reported. The listed counts also do not establish how many events occurred; the entry describes participants affected. These safety figures should be read as the registry’s posted counts, not as a complete account of event types, timing, or causality.

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

The record establishes the intended measurement question: estimate the mean relative size of discrepancies between Enlite 3 sensor readings and YSI plasma glucose values during specified testing days across a 168-hour (7-day) use period. It also identifies the connected devices, the calibration description, and the intention to pool testing-day results. These details define the target of the evaluation.

However, the registry record does not report the numerical MARD result, its standard deviation, a confidence interval, or any inferential analysis. The posted information therefore does not show how large the average discrepancy was or how precisely it was estimated. It cannot support a quantitative statement about the observed accuracy. Nor can it establish whether the measurement summary differed across days, participants, or randomized groups.

The randomized design does not change that interpretation. The primary outcome description says the two groups’ data were collected as a whole and were not intended to be analyzed separately. The record gives different FST schedules for the groups, but does not report an estimated contrast between them. It should not be described as showing a difference between schedules or as establishing a treatment effect.

In general, a pooled mean accuracy measure is not a guarantee about every reading. It summarizes the set of observations included in the analysis, while leaving other questions—such as the distribution of errors, individual-level performance, and day-specific patterns—to other statistics or analyses. Those analyses and results are not reported in the registry record.

The posted safety counts are separate from the accuracy endpoint. The registry lists zero participants affected by serious adverse events among 89 at risk and five affected by other adverse events among 89 at risk. The record does not report the death value or provide the details needed to interpret event timing and context. Taken together, the record communicates the study’s purpose and some design and safety information, but does not post enough outcome data to determine the magnitude or precision of the primary accuracy estimate.

13. Why This Trial Matters Statistically

This study illustrates a distinction between a clearly defined endpoint and a reported result. The registry record explains how the primary measure was intended to be formed: compare sensor values with YSI plasma glucose values, calculate relative absolute differences, and average them. It also identifies the testing days and says those observations were pooled. A well-described measurement target helps readers understand what an eventual estimate would mean.

At the same time, design labels do not replace outcome reporting. Randomization and a parallel-group structure are recorded, but the primary outcome description says the groups were not intended to be analyzed separately. A reader should not infer a randomized comparison simply from the design label. The target is a pooled accuracy summary, and the registry does not report its numerical value.

The record also shows why sample size and precision are different questions. Enrollment is listed, and planning assumptions are described, but no primary estimate, standard deviation, confidence interval, or numerical planning illustration is reported. In general, enrollment alone does not determine how precise a particular estimate is; the available outcome data and analysis matter. Planning illustrations are conditional calculations, not observed evidence about an outcome.

For clinical researchers, the case underscores the value of reporting the estimate together with measures of variability and uncertainty, clarifying how repeated observations are handled, and specifying which participants contribute to each analysis. Pooled results can answer an overall question while leaving patterns across time points or groups unresolved. The registry record’s missing numerical outcomes make that distinction especially important: a defined study aim is not the same as an interpretable posted finding.

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.

The main limitation for interpreting this record is that numerical primary outcome results are not reported. The MARD estimate and standard deviation are absent, as are a confidence interval and inferential analysis. As a result, the size and precision of the observed measurement discrepancy cannot be evaluated from the posted results. Numerical secondary outcome results are also not reported.

The primary outcome description says results from Days 1, 3 and 7 were pooled. No separate day-level estimates are reported, so the record does not show whether the measure varied over the stated use period. Although the randomized groups had different FST schedules, the record says the primary data were collected as a whole and were not intended for separate group analysis. No group-specific estimates are reported.

The registry record does not report how many participants or measurements contributed to the primary endpoint, how missing readings were handled, whether observations were excluded, or which analysis population was used. It does not describe a statistical method for repeated measurements or provide details of the reference measurement procedures. The relationship between the enrollment of 118 and the 89 participants listed at risk in the safety entry is not explained.

The record also does not report a sample-size justification, a statistical analysis plan, multiplicity handling, or the rationale for the groups’ different FST schedules. It lists retrospective calibration-related secondary outcomes, but their numerical findings are absent. These gaps do not establish that particular methods were or were not used; they mean those methods are not reported in the registry record.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Adult Accuracy Study of the Enlite 3 Glucose Sensor — NCT02246582.
• Registry record last updated 2016-11-02; status last verified 2016-10.
• Results first posted 2016-11-02.
• 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

The registry record describes a completed, randomized, unmasked, parallel-group evaluation of the Enlite 3 sensor. Enrollment is listed as 118; the study summary identifies participants aged 14–75 years with type 1 or type 2 diabetes. The sensor was evaluated with the Guardian Mobile App and 640G Pump over 168 hours (7 days).

The primary outcome was mean absolute relative difference between sensor values and YSI plasma glucose values during frequent sample testing on Days 1, 3 and 7. Those days’ results were to be pooled. Although participants were randomized to groups with different FST schedules, the primary outcome description says the data were collected as a whole rather than analyzed separately by group.

The registry record does not report numerical primary or secondary outcome results or analyses. It lists zero participants affected by serious adverse events among 89 at risk and five affected by other adverse events among 89 at risk. The death value is not reported.