Clinical Assessment of a Closed-loop Insulin Delivery System: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Clinical Assessment of a Closed-loop Insulin Delivery System (NCT01534013), a non-randomized na study. The primary outcome is Percentage Time in Euglycaemia.

2. Clinical Question

The study’s stated purpose is to assess the safety and efficacy of the Imperial College closed-loop insulin delivery system in subjects with type 1 diabetes. Its primary outcome is percentage time in euglycaemia, calculated from interstitial and venous blood glucose measurements during visits in which the closed-loop system is used.

The record also lists an open-loop control visit, but it does not provide a randomized assignment or describe the relationship between visits and treatment order. It therefore does not support treating the open-loop observations as a randomized, like-for-like control group. The registry record does not report a numerical primary-outcome result, a comparison estimate, or an analysis of the endpoint. The clinical question is thus identifiable from the protocol description, while the magnitude and precision of any observed difference are not reported in the registry record.

3. Trial Design

This is a completed, non-randomized study with a single-group model. The registry record lists enrollment of 23 and the condition as type 1 diabetes mellitus. The study is marked as having no masking, and its phase is listed as NA. Imperial College London is named as the sponsor. The recorded start is 2011-08 and primary completion is 2014-08.

The intervention descriptions distinguish closed-loop insulin delivery from an open-loop control visit. In the closed-loop description, a bio-inspired artificial pancreas device, subcutaneous glucose monitor and insulin pump are applied to participants with type 1 diabetes. The open-loop description names a subcutaneous glucose monitor and pump. The registry record does not report randomized allocation, and it explicitly cautions that the arms are not a like-for-like control comparison.

The primary endpoint is percentage time in euglycaemia, assessed from glucose readings during subject visits 3, 4 and 5 when insulin is being delivered using the closed-loop system. Interstitial blood glucose is to be measured every 5 minutes and venous blood glucose every 15 minutes. The stated endpoint timeframe is 18 months. The record does not explain how that timeframe relates to the visits or how the visits were scheduled.

Secondary outcomes listed in the registry include time in hypoglycaemia and hyperglycaemia, glycaemic variability, glycaemic risk, closed-loop error grid analysis, glucose area under the curve, and insulin requirement. Although the record lists eight outcome measures, it lists no analyses. It does not report a statistical analysis plan, a sample-size justification, or numerical 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

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 Percentage Time in Euglycaemia, measured over 18 months.

The primary endpoint is percentage time in euglycaemia. It is intended to describe the proportion of monitored time during which glucose values fall within the study’s euglycaemia range. The registry record says the value is to be calculated using interstitial and venous blood glucose measurements collected during subject visits 3, 4 and 5 while insulin is delivered with the closed-loop system.

Interstitial blood glucose is scheduled to be measured every 5 minutes, and venous blood glucose every 15 minutes. Frequent measurements can describe glucose patterns across a monitored visit rather than relying on a single reading. In general, a time-in-range endpoint expresses how much monitored time falls in a defined glucose category, so its interpretation depends on the definitions of the categories and the handling of observations over time.

The registry record does not report the numerical boundaries for euglycaemia, the rule for combining the two glucose measurement sources, or the calculation details for converting measurements into percentage time. Those elements are important for understanding exactly what the endpoint represents and for reproducing its calculation. No numerical primary-endpoint result, estimate of uncertainty, or comparison is reported in the registry record. The endpoint’s purpose is described, but its observed value and precision cannot be assessed from the record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The statistical information in the registry record is limited. It identifies the design as non-randomized and single-group, describes the primary endpoint, and lists secondary outcomes. It does not report an analysis plan, numerical outcome analyses, or a statistical estimate for the primary endpoint. The concepts below are general explanations of how such analyses are usually interpreted; they are not claims about methods used in this study.

What does a primary endpoint estimate represent?

Percentage time in euglycaemia is a summary of repeated glucose measurements. In general, a percentage-time measure depends on which readings qualify as euglycaemic, the monitoring interval, and how missing or invalid readings are handled. The record specifies interstitial measurements every 5 minutes and venous measurements every 15 minutes during visits 3, 4 and 5 while the closed-loop system is in use. It does not state the glucose boundaries, the method for combining measurement sources, or the rules for incomplete monitoring.

If a study reports an average percentage across participants, that average describes a participant-level summary and is not necessarily the same as pooling all glucose readings across all participants. The registry record does not say which summary was planned or used. It also does not report the number of participants contributing primary-endpoint data. Without those details and an outcome estimate, the endpoint’s distribution and precision cannot be evaluated.

Why does the allocation design matter?

In general, random allocation helps balance measured and unmeasured participant characteristics across treatment groups in expectation. A non-randomized design does not have that protection. Differences between treatment settings can therefore be associated with differences in participants, visit conditions, timing, or other factors. Statistical adjustment may address some measured differences, but it cannot automatically remove unmeasured differences. The registry record identifies the study as non-randomized and specifically states that its arms are not a like-for-like control comparison. It does not report an adjustment strategy or the order in which participants received the study conditions.

The record’s single-group model label and its descriptions of closed-loop and open-loop visits are both relevant, but do not establish a particular repeated-measures or crossover analysis. Such an analysis would depend on the actual participant-level sequence and the planned model. Those details are not reported in the registry record.

How should uncertainty be assessed?

In general, an estimate describes the observed effect or outcome, while a confidence interval describes uncertainty around that estimate under the model assumptions. A narrow interval indicates greater precision than a wide interval, but neither interval width nor statistical significance establishes that a result is free from bias. A p-value is a measure calculated under a statistical model and null hypothesis; it is not the probability that the null hypothesis is true. The registry record reports no estimate, confidence interval, standard error, or p-value for the primary endpoint. There is therefore no inferential result to interpret.

General modeling choices also depend on the outcome distribution and dependence among observations. Repeated glucose readings from the same person are not generally independent measurements. A suitable analysis may need to account for within-person correlation, but the registry record does not describe a model or variance method. If outcome variability exceeds what a chosen model assumes, over-dispersion can make standard errors too small and intervals too narrow. This is a general concern, not a finding about this study; the record contains no analysis from which to assess it.

What do the sample-size assumptions mean?

The registry record describes a planning illustration using a two-sided significance level of 0.05, equal allocation between arms, and a normal approximation. It says the effect to detect is a stated fraction of the observed effect; for time-to-event calculations, it describes total events, and for other outcomes, participants per arm. These are planning assumptions, not a description of the trial’s actual allocation or observed results. The sample-size illustrations are not observed power.

The record does not report a numerical observed effect for the primary endpoint, a sample-size justification for the trial, or the basis for selecting a target effect. Thus the planning assumptions cannot be interpreted as evidence that the study was adequately powered, nor can they establish the precision of an unreported endpoint result. Planning calculations and observed analyses answer different questions: planning explores what a design might require under assumptions, whereas an analysis describes the data collected.

10. Multiplicity, Missing Data and Analysis Populations

The registry record lists the primary outcome, percentage time in euglycaemia, and secondary outcomes covering time in hypoglycaemia, time in hyperglycaemia, glycaemic variability, glycaemic risk, closed-loop error grid analysis, glucose area under the curve, and insulin requirement. It lists eight outcome measures in total. It does not report a statistical analysis plan or any analyses.

When a study evaluates multiple outcomes or makes multiple comparisons, the chance of obtaining at least one apparently unusual result can increase. General approaches to multiplicity include defining a primary analysis in advance, organizing outcomes in a testing hierarchy, or adjusting inference across tests. The registry record does not report whether any such approach was planned or used. Since no numerical outcome analyses are listed, the record does not permit assessment of how multiplicity was handled or how it affects any claimed statistical significance.

Missing-data handling is also not reported in the registry record. The description specifies glucose measurement intervals and visits, but does not say how interrupted monitoring, unavailable values, or participants without complete visit measurements would be treated. The record does not report the number of participants contributing data to the primary or secondary outcomes, the extent of missing observations, or any sensitivity analysis. These omissions matter because the amount and pattern of missing data can affect the interpretation of repeated-measurement outcomes.

The analysis populations are not described. The record does not say whether the primary outcome was analyzed for all enrolled participants, only those with measurements at specified visits, or another defined group. No participant flow or exclusion criteria for an analysis population are reported. As a result, enrollment of 23 should not be treated as the denominator for a particular outcome analysis without further information.

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.

The registry record lists 12 participants at risk in the closed-loop insulin delivery group and 12 at risk in the open-loop group. For each group, it records 0 participants affected by serious events and 0 affected by other events. No serious-event terms are listed. The death count is not reported in the record.

These entries are counts as recorded in the registry. They do not provide event-level descriptions, timing, duration, severity grading, or an explanation of how events were collected and classified. The record also does not explain why the listed participants at risk differ from the enrollment figure of 23. The denominators should therefore be read as the safety entries for the respective groups, not as a complete account of participant flow.

In general, a recorded count of zero means that no affected participants were entered for that category in the available record; it does not prove that no events occurred outside the definitions, observation period, or reporting process. Without further information on follow-up, ascertainment, and the death fields, the safety entries cannot establish the absence of risk or support a detailed comparison between the study conditions.

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

The registry record identifies the study question, intervention descriptions, primary endpoint, and selected safety counts. It states that percentage time in euglycaemia is to be calculated from glucose monitoring during visits 3, 4 and 5 when the closed-loop system is used. It also records 0 participants affected by serious events and 0 affected by other events in each listed group, with 12 participants at risk in each.

Those descriptive entries do not provide a numerical primary-outcome result. The registry record lists no analyses and does not report an estimate, uncertainty interval, or hypothesis test for percentage time in euglycaemia. It therefore cannot show the size or precision of an observed glucose outcome, or whether one treatment setting differed from the other on that endpoint.

The non-randomized design is an important boundary on interpretation. The record explicitly cautions that the arms are not a like-for-like control comparison. An observed contrast, if reported elsewhere, would need to be interpreted in light of the allocation and visit structure, which are not fully described in the registry record. A contrast would not by itself establish that the treatment condition caused the difference.

In general, the absence of a reported result is not evidence that the outcome was unchanged, and a recorded zero in an adverse-event category is not proof that risk is absent. The record supports describing what was planned and what was entered in the safety fields; it does not support a treatment recommendation or a claim that the device does or does not work.

13. Why This Trial Matters Statistically

This study illustrates why a clearly described endpoint is only part of the statistical evidence needed to interpret a clinical trial. The registry record specifies percentage time in euglycaemia and gives the glucose measurement schedule: interstitial readings every 5 minutes and venous readings every 15 minutes during visits 3, 4 and 5. That description identifies the intended measurement, but a numerical estimate and information about uncertainty are needed to assess the observed outcome.

The design also matters. The record labels the study non-randomized and single-group, and warns that the listed arms are not a like-for-like control comparison. This makes it important to distinguish treatment-condition descriptions from a randomized estimate of treatment effect. In general, statistical interpretation must consider how participants and visits were assigned, what observations were included, and whether the analysis accounts for the structure of repeated measurements. Those details are not reported in the registry record.

The record further shows why registry entries should be read as distinct pieces of evidence. Enrollment is listed as 23, while the safety record gives 12 participants at risk in each group. The record does not reconcile these figures. It lists outcomes and safety counts, but no analyses. Planning assumptions, where recorded, do not substitute for observed effect estimates or an analysis of precision.

For biostatisticians and clinical researchers, the central statistical lesson is that a protocol description and safety count do not establish an efficacy result. Interpretation requires transparent reporting of outcome estimates, uncertainty, analysis populations, missing-data handling, and the statistical approach. The registry record’s omissions set limits on the conclusions that can be drawn from its entries.

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 trial from the registry record is the absence of numerical outcome results and analyses. Although percentage time in euglycaemia is designated as the primary outcome, no observed value, comparison estimate, measure of uncertainty, or p-value is reported. The listed secondary outcomes likewise have no numerical analyses in the record.

The design is non-randomized, and the registry explicitly cautions that the arms are not a like-for-like control comparison. Allocation details, treatment order, and the relationship between the closed-loop and open-loop visits are not reported. The record therefore does not establish whether the groups or visits can be compared directly, or how potential differences were addressed.

Important endpoint details are also absent. The registry record does not give the numerical glucose boundaries defining euglycaemia, the method for combining interstitial and venous readings, or the rules for handling missing measurements. It does not report how many participants contributed to each outcome, how analysis populations were defined, or whether incomplete visits affected inclusion.

The enrollment figure of 23 and the safety denominators of 12 per group are not reconciled. The record does not provide participant flow, event descriptions, or a numerical death count. It also does not report a sample-size justification, a statistical analysis plan, multiplicity procedures, or missing-data sensitivity analyses. These gaps prevent detailed assessment of both efficacy and safety, and they limit reproducibility of any analysis beyond the measurement description recorded in the registry.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Clinical Assessment of a Closed-loop Insulin Delivery System — NCT01534013.
• Registry record last updated 2020-10-30; status last verified 2020-10.
• Results first posted 2019-08-28.
• 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, non-randomized, single-group study sponsored by Imperial College London, with enrollment of 23 participants with type 1 diabetes mellitus. Its stated purpose is to assess the safety and efficacy of the Imperial College closed-loop insulin delivery system.

The primary outcome is percentage time in euglycaemia, based on glucose measurements during visits 3, 4 and 5 while using the closed-loop system. The record also lists secondary glucose and insulin outcomes. It identifies an open-loop control visit but cautions that the arms are not a like-for-like control comparison.

No numerical outcome analyses are reported. The safety entries list 12 participants at risk and 0 affected by serious events and 0 affected by other events in each group; a death count is not reported. The registry record does not report an analysis plan, participant-level outcome denominators, or the handling of missing data.