This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial results on this page are restricted to the results posted in the ClinicalTrials.gov record; calculations added by Clinical Biostats are labelled as such.
1. Trial at a Glance
This page covers the registry results for PRECISION Study: Multi-center Study of Performance of a Novel Implanted CGM System Using a Next Generation Transmitter and Algorithm (NCT03251079), a na na study. The primary outcome is Accuracy of CGM System Through 90 Days Post-Insertion (MARD for Paired CGM and Reference Glucose Measurements).
| Feature | Registry value |
|---|---|
| ClinicalTrials.gov identifier | NCT03251079 |
| Phase | NA |
| Condition | Diabetes Mellitus, Diabetes Mellitus, Type 1, Diabetes Mellitus, Type 2 |
| Allocation | Not applicable (single group) |
| Design model | Single Group |
| Masking | None (open label) |
| Enrollment | 36 |
| Arms in the primary outcome | n/a |
| Lead sponsor | Senseonics, Inc. |
| Results posted | Yes |
2. Clinical Question
Population
The registry record describes a prospective, multi-center study in people with diabetes mellitus, including type 1 and type 2 diabetes. Enrollment was 36. Further participant characteristics, including age and baseline clinical details, are not reported in the registry record.
Intervention
Participants used the Senseonics continuous glucose monitoring system, an implanted sensor system paired with a next generation transmitter and algorithm. The record identifies the intervention as a device. It does not describe additional treatment components or specify details of sensor insertion procedures.
Comparator
The study compared CGM readings with paired measurements from a Yellow Springs Instrument glucose analyzer, described as the reference standard. This is a comparison of sensor readings with reference measurements, not a randomized comparison between treatment groups. The registry lists a single-group design and no allocation.
Primary question
The primary question was how closely the CGM system’s glucose readings agreed with paired YSI reference readings through 90 days of sensor use. The endpoint was mean absolute relative difference, or MARD, expressed as a percentage. The registry record does not report the observed MARD estimate or its 95% confidence interval, so the magnitude and precision of the measured agreement cannot be assessed from the record.
The study also states that it evaluated the effects of compression on performance and the safety of the system. The registry record does not provide details of the compression assessment in the material describing the primary endpoint. The primary accuracy question is therefore distinct from safety and from any other performance questions.
3. Trial Design
The PRECISION Study is described as a prospective, multi-center evaluation of an implanted continuous glucose sensor using a next generation transmitter and algorithm. The registry lists the design as single-group, with allocation recorded as not applicable and masking as none. There is no randomized control arm. Instead, the accuracy endpoint uses paired CGM and YSI reference glucose readings for comparison.
The registry reports enrollment of 36. The study start date is 2017-07-25, primary completion is dated 2018-02-01, and the status is completed. These dates describe the registry record; they do not indicate when individual participants began or completed sensor use.
The intervention is identified as a continuous glucose monitoring device, specifically the Senseonics continuous glucose monitoring system. The study’s stated purpose was to evaluate system performance compared with YSI reference measurements, examine effects of compression on performance, and assess safety. The record does not report further details about the study procedures, eligibility criteria, site locations, or how often measurements were collected.
Because the design is single-group, the study is not a randomized comparison of the device against another treatment or monitoring strategy. The YSI analyzer serves as a measurement reference for the accuracy assessment, rather than as a separate participant group. Any interpretation of the primary endpoint should therefore focus on agreement between paired readings and on the uncertainty around that estimate, not on a between-group treatment effect.
The registry does not report a protocol-level statistical analysis plan, sample-size justification, or interim-analysis plan. Those details cannot be inferred from the prospective design, the enrollment figure, or the fact that the study is marked completed.
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
| Stage | Continuous Glucose Monitoring Device |
|---|---|
| STARTED | 36 |
| COMPLETED | 35 |
| NOT COMPLETED | 1 |
Reasons for not completing this period:
| Reason | Continuous Glucose Monitoring Device |
|---|---|
| Withdrawal by Subject | 1 |
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.
| Characteristic | Unit | Continuous Glucose Monitoring Device |
|---|---|---|
| Participants analysed | 35 | |
| Age, Continuous | years | 51.6 (SD 15.7) |
| Sex: Female, Male: Female | Participants | 17 |
| Sex: Female, Male: Male | Participants | 18 |
| Race/Ethnicity, Customized: Caucasian | Participants | 32 |
| Race/Ethnicity, Customized: Black or African American | Participants | 1 |
| Race/Ethnicity, Customized: Asian | Participants | 2 |
| Race/Ethnicity, Customized: American Indian or Alaska Native | Participants | 0 |
| Race/Ethnicity, Customized: Native Hawaiian or Other Pacific Islander | Participants | 0 |
| Race/Ethnicity, Customized: More than One Race | Participants | 0 |
| Race/Ethnicity, Customized: Hispanic | Participants | 4 |
| Race/Ethnicity, Customized: Non-Hispanic | Participants | 31 |
| Region of Enrollment: United States | participants | 35 |
| Body Mass Index Class: Normal, Overweight, Obese | kg/m2 | 28.2 (SD 5.4) |
| Dominant Hand: Right, Left: Right | participants | 33 |
| Dominant Hand: Right, Left: Left | participants | 2 |
| Diabetes Type: Type I, Type II: Type I | Participants | 25 |
| Diabetes Type: Type I, Type II: Type II | Participants | 10 |
| Years Since Diabetes Diagnosis | years | 26.0 (SD 14.3) |
| Type of insulin therapy: None | Participants | 5 |
| Type of insulin therapy: Multiple daily injections | Participants | 11 |
| Type of insulin therapy: Continuous insulin infusion pump | Participants | 19 |
| History of ketoacidosis and hypoglycemia in past 6 months: Ketoacidosis | Participants | 0 |
| History of ketoacidosis and hypoglycemia in past 6 months: Hypoglycemia | Participants | 0 |
6. Primary Endpoint
The registry lists the primary outcome as Accuracy of CGM System Through 90 Days Post-Insertion (MARD for Paired CGM and Reference Glucose Measurements), measured over 90 days.
| Element | Registry entry |
|---|---|
| Result type | Mean |
| Measure of spread | 95% Confidence Interval |
| Unit | percent |
| Endpoint classification | Continuous Endpoint |
The primary endpoint is the mean absolute relative difference, or MARD, for paired Senseonics CGM and reference glucose measurements through 90 days of sensor use. It is expressed as a percentage. For each pair, the absolute difference between the sensor reading and reference reading is scaled relative to the reference reading; MARD averages those relative differences across the included pairs. The endpoint description states that lower MARD values indicate higher accuracy.
MARD is used to describe the size of sensor-reading errors relative to the reference values, rather than only the size of errors in the glucose measurement’s original units. This makes the endpoint a summary of relative agreement across the included paired readings. It does not, by itself, show whether errors are consistently in one direction, how error varies across glucose ranges, or how the device performs for particular participants or circumstances.
The registry specifies a mean and a 95% confidence interval as the summary measure. However, the observed MARD and its confidence interval are not reported in the registry record. The endpoint definition explains what was intended to be measured, but without the estimate and interval it does not establish the observed level of accuracy or how precisely it was estimated.
7. Why No Re-analysis Is Shown
8. Statistical Interpretation
The registry identifies MARD through 90 days of sensor use as the primary accuracy endpoint and explains how it is calculated. It does not report the observed MARD value or its 95% confidence interval. As a result, the posted registry information does not allow readers to quantify the average relative difference between CGM readings and YSI reference readings, or to judge the precision of that estimate. No conclusion about the magnitude of the observed accuracy result can be drawn from the endpoint definition alone.
The design helps define what kind of evidence the study could provide. Sensor readings were assessed against paired reference measurements within a single-group study. The YSI analyzer is a measurement comparator, not a separate randomized group. The study therefore addresses agreement with a reference standard, rather than a causal comparison between two assigned treatments. Its design does not by itself establish how results compare with another device or monitoring approach.
The record reports enrollment of 36, while the safety information lists 35 people at risk in the device group. These figures refer to different registry entries and should not be combined or treated as interchangeable denominators. The record does not explain the difference. It also does not provide participant flow or the number of paired readings included in the primary endpoint, so the amount of endpoint information contributing to MARD is not reported.
The listed safety figures include 3 people with serious adverse events, 15 with other adverse events, and 0 deaths in the device group; the listed at-risk count is 35. These are descriptive registry figures. They do not, by themselves, establish the cause of an event or how event risk compares with another device or with no device. The registry record does not provide enough information to evaluate event timing, exposure-adjusted rates, or the relationship of individual events to device use.
Statistical precision would ordinarily be assessed using the confidence interval around the MARD estimate. In general, a confidence interval describes uncertainty in an estimate under the analysis assumptions; it does not guarantee that the estimate captures every source of uncertainty, such as missing measurements or limitations in the analysis population. Since the interval is not reported, the precision of the primary endpoint cannot be assessed.
The record lists 2 outcome measures but gives details for the primary MARD endpoint only. It lists no analysis entries. The analysis population, handling of missing readings, and procedures for evaluating repeated measurements are not reported in the registry record. These omissions limit interpretation of how the endpoint was formed and how representative it is of the participants and readings collected.
9. Statistical Methods Explained
The primary endpoint is a continuous measure: MARD, summarized as a mean with a 95% confidence interval. The following concepts are general statistical explanations; they do not establish which methods were used beyond what the registry record describes.
What does MARD summarize?
MARD expresses each paired sensor-reference difference relative to the reference reading, then averages those relative differences. Taking the absolute difference removes the direction of error: a sensor reading above the reference and one below it both contribute a positive error magnitude. The average therefore conveys the typical size of relative discrepancies across the included pairs, but it does not indicate whether readings tend to be systematically high or low.
Because the error is scaled to the reference value, the same absolute difference can contribute differently depending on the reference reading. This relative scaling is useful when a measure should reflect error in proportion to the measured value. It also means the distribution of reference glucose values can matter to the overall summary. The registry defines the calculation but does not report the distribution of reference values, MARD by glucose range, or other summaries that might help characterize performance across the measurement range.
Why report a confidence interval?
In general, a confidence interval communicates statistical uncertainty around an estimate. A narrower interval indicates greater precision under the assumptions of the analysis; a wider interval indicates less precision. The registry specifies a 95% confidence interval for the mean MARD, but it does not report the interval’s limits. Neither the precision nor the range of values compatible with the result can therefore be evaluated from the registry record.
A confidence interval is not a measure of every possible uncertainty. It generally reflects sampling variability under a particular analysis model and depends on how observations are treated. If paired measurements are collected repeatedly from the same person, those measurements may be correlated. Treating correlated readings as if each were independent can understate uncertainty. An analysis may need to account for clustering or repeated observations; the statistical approach used for this endpoint is not reported in the registry record.
Why do paired measurements matter?
Pairing connects each CGM reading with a reference reading used for comparison. The MARD definition operates on these pairs, so the number and selection of pairs influence the calculated summary. In general, readings from the same participant can share characteristics, and a participant contributing many readings may contribute more to a reading-level average than someone contributing fewer readings. Whether the study’s analysis accounted for this structure, and whether it summarized results at the reading or participant level, is not reported.
The endpoint description refers to the total number of CGM and reference pairs after 90 days of sensor use, but the registry record does not report that number. It also does not report how readings were selected, whether any pairs were excluded, or how missing or invalid readings were handled. These details matter because a summary based on available pairs can be influenced by which pairs are observed and included.
What can a single-group reference comparison establish?
In general, comparing a device reading with a reference measurement can characterize agreement with that reference under the study’s measurement conditions. It is not equivalent to a randomized treatment comparison. A single-group design does not separate device performance from every potential influence on measurements, and it cannot estimate a contrast against a separately assigned control group. The registry identifies YSI as the reference standard, but does not report enough procedural detail to assess all conditions under which paired readings were obtained.
The record states that compression effects on performance were also evaluated, but the primary endpoint description does not explain the methods or results for that assessment. No subgroup or compression-specific analysis is reported in the registry record. Thus, no particular influence of compression can be inferred from the primary endpoint description.
How should planning illustrations be read?
In general, sample-size calculations based on an observed effect are planning illustrations, not evidence that a study had a particular level of power after completion. The stated illustration assumptions are a two-sided significance level of 0.05, equal allocation between two arms, a normal approximation, and an effect set to a stated fraction of the observed effect. Those assumptions describe a hypothetical two-arm planning exercise and do not match the registry’s single-group design as a direct account of its design. The record does not report the trial’s own sample-size justification. No planning illustration should be read as observed power or as proof that the study was adequately sized.
More broadly, the registry specifies the endpoint and its intended summary but does not provide the numerical MARD result, uncertainty interval, or analysis details needed to reproduce or fully evaluate the primary analysis. The definitions explain the target of measurement; they do not substitute for reported results.
10. Multiplicity, Missing Data and Analysis Populations
The registry identifies MARD through 90 days as the primary endpoint and lists 2 outcome measures. It does not provide details for the additional outcome measure in the material describing the primary endpoint. The registry record lists no analysis entries, and a statistical analysis plan is not reported. Consequently, it is not possible to determine from the record which analyses were prespecified, whether multiple comparisons were planned, or whether any adjustment for multiplicity was used.
Multiplicity can arise when several endpoints, time points, subgroups, or statistical tests are examined. In general, conducting many tests can increase the chance of finding an apparently unusual result even when there is no corresponding underlying difference. This consideration does not establish that the study conducted multiple tests or that an adjustment was needed; those details are not reported. The primary endpoint designation identifies the principal accuracy measure, but does not reveal how any other analyses were handled.
The handling of missing data and incomplete measurement pairs is also not reported in the registry record. The MARD definition refers to paired sensor and reference readings, but the record does not state how missing, invalid, or unmatched readings were treated. It does not report the number of usable pairs, the number of participants contributing to the primary analysis, or whether an analysis population was defined by a specific inclusion rule.
These omissions matter because an analysis based only on available pairs can differ from one using other assumptions about unavailable measurements. In general, missingness can affect both the estimate and its uncertainty, particularly if the likelihood of missing readings is related to the measurement or participant characteristics. No assumptions about the reason for missing data or the method used to address them can be made from the registry record.
11. Safety Results
The registry reports serious adverse events as affected participants divided by participants at risk.
| Arm | Participants with serious adverse events | Participants at risk | Percentage |
|---|---|---|---|
| Continuous Glucose Monitoring Device | 3 | 35 | 8.6% |
Deaths
| Arm | Deaths (participants affected of at risk) |
|---|---|
| Continuous Glucose Monitoring Device | 0 of 35 |
Other (non-serious) adverse events
Participants with at least one non-serious event at or above the registry's 0% reporting threshold.
| Arm | Participants affected of at risk |
|---|---|
| Continuous Glucose Monitoring Device | 15 of 35 |
Most frequent serious adverse event terms
| Event term | Organ system | Continuous Glucose Monitoring Device |
|---|---|---|
| Other - Gastroenteritis | Gastrointestinal disorders | 1 of 35 |
| Hypoglycemic episode | Endocrine disorders | 1 of 35 |
| Other - Wound infection left foot, grade 4 | Musculoskeletal and connective tissue disorders | 1 of 35 |
Most frequent non-serious adverse event terms
| Event term | Organ system | Continuous Glucose Monitoring Device |
|---|---|---|
| Common Cold | Respiratory, thoracic and mediastinal disorders | 4 of 35 |
| Headache | Nervous system disorders | 4 of 35 |
| Other - Upper Respiratory Infection | Respiratory, thoracic and mediastinal disorders | 3 of 35 |
| Sensor Location Site - Pain/Discomfort | Skin and subcutaneous tissue disorders | 2 of 35 |
| Other - Pneumonia | Respiratory, thoracic and mediastinal disorders | 1 of 35 |
| Other - Pilonidal cyst | Skin and subcutaneous tissue disorders | 1 of 35 |
| Other - Difficulty removing sensor | Skin and subcutaneous tissue disorders | 1 of 35 |
| Adhesive Patch Location Site - Irritation including redness, excoriation or ulceration | Skin and subcutaneous tissue disorders | 1 of 35 |
| Other - Darkening of skin at proximal end of sensor | Skin and subcutaneous tissue disorders | 1 of 35 |
| Other - Vaginal Infection | Reproductive system and breast disorders | 1 of 35 |
Serious events with the highest arm-level rate: other - gastroenteritis (2.9%), hypoglycemic episode (2.9%), other - wound infection left foot, grade 4 (2.9%).
The registry’s device group lists 35 people at risk for the safety summaries. It records serious adverse events in 3 people, other adverse events in 15 people, and 0 deaths. These figures are descriptive counts in the registry record. They are not evidence, by themselves, that the device caused an event or that an event was unrelated to it. The record does not provide timing, event-level details, or a comparison group for estimating relative risk.
The listed serious-event terms include Other - Gastroenteritis, Hypoglycemic episode, and Other - Wound infection left foot, grade 4, each reported as 2.9%. These are the top serious terms named in the record. The registry does not explain whether categories could overlap or provide additional detail about event duration, severity assessment, or attribution. The listed percentages should be read as reported, not recalculated into a different denominator or interpreted as comparative rates.
The safety count at risk differs from the enrollment figure of 36. The registry record does not explain that difference. Serious-event and other-event counts should also be kept distinct: the record presents them in separate categories, and it does not describe whether the same person could appear in both. The absence of reported deaths does not address other aspects of safety or establish a comparison with another monitoring system.
12. What the Result Does, and Does Not, Mean
The study was designed to evaluate an implanted Senseonics CGM system by comparing paired sensor readings with YSI reference glucose measurements through 90 days of sensor use. Its primary endpoint, MARD, is defined as the average relative size of the absolute differences. The registry also states that lower MARD values indicate higher accuracy. These points describe the endpoint and intended comparison, not the observed result.
The observed MARD and its 95% confidence interval are not reported in the registry record. Therefore, the record does not show how large the average relative differences were or how precisely they were estimated. Without those values, it is not possible to characterize the measured accuracy result from the registry information. An endpoint definition is not a substitute for an endpoint estimate.
The study’s single-group design means that YSI readings function as a reference standard rather than as a separate control group. The study can be understood as an assessment of agreement with that reference, subject to the measurement procedures and analysis used. It does not provide a randomized comparison between the CGM system and another treatment or monitoring strategy. The record does not establish comparative effectiveness or a causal effect relative to another device.
The safety entries describe adverse-event counts and listed serious-event terms for the device group. They should not be interpreted as establishing causation, comparative risk, or the absence of safety concerns beyond the reported entries. The registry does not report the detailed event context needed for those assessments.
In general, statistical estimates require consideration of both the observed value and its uncertainty, as well as how measurements and missing observations were handled. The registry record does not report the primary estimate, interval, measurement-pair count, or analysis population. The conclusions that can be drawn are consequently limited to the study’s stated objective, endpoint definition, design, and recorded safety figures. No conclusion about whether the device does or does not work follows from the available registry information.
13. Why This Trial Matters Statistically
This study raises a central measurement question in device evaluation: how closely do sensor readings agree with a reference method when both are measured in pairs? MARD offers a compact summary of relative discrepancy, and the registry explicitly defines how the endpoint is intended to represent agreement. The endpoint therefore makes the statistical target clear even though the numerical result is not reported.
The design also illustrates why the role of a comparator must be stated precisely. YSI is the reference used to evaluate CGM readings, not a randomized participant group. That distinction determines what the analysis can address. A reference-based accuracy assessment and a treatment comparison answer different questions and rely on different interpretations.
Repeated paired readings create statistical considerations beyond the choice of a mean. In general, measurements from the same participant may be correlated, and the number of readings contributed by each participant can affect a reading-level average. Reporting the analysis unit, the number of pairs, and methods for accounting for repeated observations would help readers understand how uncertainty was estimated. Those details are not reported in the registry record.
Precision is also essential to interpretation. A mean MARD without its confidence interval would leave uncertainty unclear; in this record, neither is reported. The endpoint definition, enrollment figure, and completed status cannot replace the estimate or its uncertainty. This makes the distinction between a planned measurement and a reported result particularly important.
Finally, the registry’s safety entries are separate from the accuracy endpoint. Descriptive event counts can inform what was recorded, but they do not by themselves establish causation or comparative safety. Taken together, the record illustrates the value of reporting endpoint estimates, uncertainty, analysis methods, and event context alongside a clear design description.
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 the primary endpoint is that the registry record does not report the observed MARD or its 95% confidence interval. It therefore does not quantify the reported accuracy result or its precision. The number of paired readings used to calculate MARD is also not reported, limiting assessment of the information underlying the estimate.
The record describes a single-group study using YSI as the reference standard. It does not include a randomized comparison group, so the design does not support a direct treatment contrast against another monitoring approach. The record also does not report detailed measurement procedures, participant characteristics, eligibility criteria, or site information. These details would help readers assess the setting and generalizability of the findings.
Important analysis details are absent. The registry record does not report the statistical analysis plan, the analysis population, the handling of missing or invalid pairs, or methods for accounting for repeated observations within participants. The record lists 2 outcome measures but only describes the primary MARD endpoint in detail, and it lists no analysis entries. Multiplicity procedures and subgroup analyses are not reported.
Enrollment is reported as 36, while the safety group lists 35 at risk; the registry does not explain the difference. Safety information is presented as counts and selected serious-event terms, without event timing, attribution, or comparative rates. The record also states that compression effects were evaluated but does not describe the corresponding methods or results in the primary endpoint information.
These limitations concern what can be interpreted from the registry record; they do not establish that a procedure was or was not performed. The statistical plan, sample-size justification, and interim analyses are not reported in the registry record.
15. Related Tutorials
16. Related Statistical Calculators
17. Sources
- ClinicalTrials.gov: PRECISION Study: Multi-center Study of Performance of a Novel Implanted CGM System Using a Next Generation Transmitter and Algorithm — NCT03251079.
- Registry record last updated 2024-02-12; status last verified 2024-02.
- Results first posted 2024-02-12.
- 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.
Continue through the Clinical Biostats statistical pathway
Use this trial's endpoint as a starting point for deeper study of effect measures, confidence intervals, and sample-size planning.
18. Record Summary
The PRECISION Study, NCT03251079, is a completed, prospective, multi-center, single-group evaluation of a Senseonics implanted continuous glucose monitoring system using a next generation transmitter and algorithm. The registry reports enrollment of 36 and identifies YSI glucose analyzer measurements as the reference standard for the accuracy assessment.
The primary endpoint is mean absolute relative difference for paired CGM and reference glucose readings through 90 days of sensor use, summarized as a mean with a 95% confidence interval. The registry record does not report the observed MARD estimate or confidence interval. It lists 2 outcome measures and no analysis entries.
The safety record lists 35 people at risk, 3 with serious adverse events, 15 with other adverse events, and 0 deaths. Selected serious-event terms are reported with percentages. The record does not provide the statistical analysis plan, handling of missing readings, or details needed to interpret the primary endpoint’s analysis population and precision.