Closed-Loop Glucagon Pump for Treatment of Post-Bariatric Hypoglycemia: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Closed-Loop Glucagon Pump for Treatment of Post-Bariatric Hypoglycemia (NCT03255629), a randomized phase 1, phase 2 study. The primary outcome is Number of Participants With Meal-provoked Hypoglycemia, Defined as Sensor Glucose <65 mg/dL.

2. Clinical Question

This question concerns an outcome observed during controlled meal challenges, rather than a measure of long-term experience in everyday life. The registry summary also mentions hypoglycemia in response to exercise, but the primary endpoint described in the posted results is measured following mixed-meal challenges. The trial enrolled 18 participants, while the primary analysis population consisted of the 12 who completed both visits. The posted endpoint therefore describes results among those completers, not every person enrolled. The record does not report further details that would establish how representative those completers were of the broader population with post-bariatric hypoglycemia.

3. Trial Design

This was a randomized, quadruple-masked, crossover study listed in phases 1 and 2. In a crossover design, a participant receives each of the compared conditions at different times. Here, each participant was scheduled for two mixed-meal tolerance tests: one with glucagon and one with matched placebo. The order was randomized, so some participants received glucagon first and others placebo first. The second challenge was conducted within two weeks of the first.

The crossover arrangement makes each participant a comparison with themselves across the two conditions. That can reduce the influence of stable differences between participants, but it also means that analysis should account for the paired observations. Randomization of treatment order helps distribute possible order effects, while masking is intended to limit expectations or knowledge of treatment from influencing trial conduct or assessment. The registry does not report further details about how masking was implemented or whether it remained intact.

The registry lists 18 enrolled participants and a completed status. For the primary outcome, it states that 12 participants completed both mixed-meal visits and were included in the analysis. Thus, the posted primary comparison is based on a smaller analysis population than the enrollment count. The registry record does not report the reasons that participants were not included in the primary analysis or describe how missing observations were handled.

The intervention is identified as glucagon delivered by a closed-loop glucagon pump; the comparator is placebo/control. The arm descriptions indicate that either 1 or 2 doses could be given at a study visit depending on patient response. The registry record does not report the original sample-size justification, detailed allocation procedures, a full protocol, or the original statistical analysis plan. The study’s dates and sponsor are listed in the registry, but those administrative details do not establish the assumptions or procedures used for the statistical analyses.

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: Between September 2017 and August 2018, 23 individuals were screened and 18 enrolled (consort diagram, Fig. 2). Two participants withdrew before the first mixed-meal study visit because of inability to obtain adequate intravenous access.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: 12 Enrolled Participants Completed 2 Mixed-Meal Study Visits and Were Included in the Analysis

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: Demographic Information for 12 Enrolled Participants Who Completed 2 Mixed-Meal Study Visits and Were Included in the Analysis

6. Primary Endpoint

The registry lists the primary outcome as Number of Participants With Meal-provoked Hypoglycemia, Defined as Sensor Glucose <65 mg/dL, measured over Measured following 2 challenges: 1 with glucagon and 1 with vehicle. The second challenge was conducted within two weeks of the first..

The primary endpoint counts participants with meal-provoked hypoglycemia, defined as sensor glucose below <65 mg/dL. It was measured following two challenges, one with glucagon and one with vehicle, with the second challenge conducted within two weeks of the first. The endpoint therefore asks whether a participant experienced at least one qualifying low sensor-glucose result during the relevant challenge condition; it is a participant count, not a count of every low reading or a measure of how long glucose remained low.

A threshold-based endpoint turns sensor readings into a clinically interpretable event classification for the trial question. Comparing the number of participants meeting the definition in each phase provides a direct description of how often the specified event occurred under each condition in the analyzed group. The registry identifies prevention of meal-provoked hypoglycemia as a primary endpoint, so this outcome is central to the trial’s stated question.

There are limits to what the endpoint captures. It does not by itself describe the depth or duration of low readings, the timing of the event, or the full glucose profile during the challenge. It is also specific to the sensor threshold stated for this endpoint; other posted outcomes use different thresholds or plasma glucose. The result should therefore be read as evidence about this defined participant-level event during the study challenges, rather than as a complete account of glucose control or of longer-term outcomes.

7. Results Posted in the Registry

• Study Drug (Glucagon) Phase: 8.3% (1/12) versus 41.7% (5/12) on control; risk difference -33.3 percentage points (95% CI -65.3 to -1.4), risk ratio 0.20 (0.03–1.47); unadjusted two-proportion z-test p = 0.059.

8. Secondary Efficacy Endpoints

The registry also posts results for secondary outcomes. For each outcome below, the endpoint type, the test applied to the posted summary statistics, the result and its interpretation are given in turn. Secondary endpoints are usually not adjusted for multiple testing, so they support rather than confirm the primary finding.

Number of Participants With Meal-provoked Hypoglycemia, Defined as Sensor Glucose <60 mg/dL

• Endpoint type: Binary: the number of participants with the event, compared as proportions.
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Study Drug (Glucagon) Phase: 8.3% (1/12) versus 25.0% (3/12) on control; risk difference -16.7 percentage points (95% CI -45.7 to 12.4), risk ratio 0.33 (0.04–2.77); unadjusted two-proportion z-test p = 0.273.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Study Drug (Glucagon) Phase and the control. The interval gives the range of effects that remain compatible with the data.

Number of Participants With Hypoglycemia Rescue Administered

• Endpoint type: Binary: the number of participants with the event, compared as proportions.
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Study Drug (Glucagon) Phase: 0.0% (0/12) versus 58.3% (7/12) on control; risk difference -58.3 percentage points (95% CI -86.2 to -30.4); unadjusted two-proportion z-test p = 0.002.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the proportion with the event is statistically lower for Study Drug (Glucagon) Phase than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Number of Participants With Meal-provoked Hypoglycemia, Defined as Plasma Glucose <60 mg/dL

• Endpoint type: Binary: the number of participants with the event, compared as proportions.
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Study Drug (Glucagon) Phase: 0.0% (0/12) versus 25.0% (3/12) on control; risk difference -25.0 percentage points (95% CI -49.5 to -0.5); unadjusted two-proportion z-test p = 0.064.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Study Drug (Glucagon) Phase and the control. The interval gives the range of effects that remain compatible with the data.

The registry lists secondary outcomes spanning several definitions of hypoglycemia, rescue administration, rebound hyperglycemia, and time spent in glucose ranges. These outcomes complement the primary sensor-glucose threshold by considering different thresholds, a response to hypoglycemia, and broader descriptions of glucose during the challenges. The detailed secondary results reported in the registry are not complete for every listed outcome, so the overall pattern cannot be assessed across all measures.

Among the posted binary secondary results, the glucagon phase has fewer participants with events for the outcomes with counts available. This direction is consistent with the primary count comparison, but the outcomes are not interchangeable: they use different definitions and capture different aspects of the challenge. The registry also reports a sponsor McNemar result for rescue administration, while some secondary recalculations are based only on summary counts. As with the primary result, paired and unadjusted methods should be distinguished.

Secondary outcomes generally carry less evidential weight than a prespecified primary endpoint, particularly when many outcomes are examined and no multiplicity procedure is reported. Testing numerous outcomes creates more opportunities for apparently notable findings to occur by chance. The registry record does not report a multiplicity adjustment or a stated hierarchy for confirmatory interpretation. Secondary results can help describe the pattern of observations and generate questions, but they should not independently establish a definitive conclusion. Results for unreported secondary measures are not reported in the registry record.

Secondary outcomes registered for this trial, for reference (only those above were recalculated):

• Number of Participants With Meal-provoked Hypoglycemia, Defined as Sensor Glucose <60 mg/dL
• Number of Participants With Rebound Hyperglycemia (Defined as Glucose Levels Above 180 mg/dl).
• Number of Participants With Hypoglycemia Rescue Administered
• Number of Participants With Meal-provoked Hypoglycemia, Defined as Plasma Glucose <60 mg/dL
• Number of Participants With Meal-provoked Hypoglycemia, Defined as Plasma Glucose <55 mg/dL
• Number of Participants With Meal-provoked Hypoglycemia, Defined as Sensor Glucose <55 mg/dL
• Percent Time Plasma Glucose in Range After the Final Dose of Study Drug or Vehicle, Which Was Either 1 or 2 Doses Depending on Patient Response
• Percent Time Sensor Glucose in Range After Drug Delivery After the Final Dose of Study Drug or Vehicle, Which Was Either 1 or 2 Doses Depending on Patient Response

9. Statistical Interpretation

10. Statistical Methods Explained

The primary outcome is binary at the participant level: each participant is counted as having or not having the specified hypoglycemia event in a phase. The reported counts are 1 of 12 in the glucagon phase and 5 of 12 in the placebo/control phase. A risk difference expresses the difference between those event proportions, with the comparison oriented as glucagon minus control. A negative estimate therefore corresponds to a lower observed event proportion in the glucagon phase in this sample. The posted estimate is -0.333.

A confidence interval gives a range of values compatible with the data under the method’s assumptions. It is not a range containing a known treatment effect with certainty, nor does it say that every value in the interval is equally plausible. The primary unadjusted interval runs from -0.653 to -0.013. Because it does not include zero, it indicates that the summary-statistic interval is consistent with a negative risk difference at the stated confidence level. Its breadth also signals uncertainty about the size of that difference.

The sponsor reports a McNemar analysis with p=0.103. McNemar’s test is designed for paired binary observations, such as the same participants observed under two conditions. It focuses on participants whose outcome differs between conditions: the paired discordant responses carry the information about whether one condition has more events than the other. Participants with the event in both phases, or with no event in either phase, do not contribute to the direction of the paired contrast in the same way. The aggregate counts alone do not reveal the discordant-pair pattern, so the sponsor’s paired test cannot be reproduced from the margins presented in the record.

A p-value measures how surprising data at least as incompatible with a specified null hypothesis would be if that null model and its assumptions held. It is not the probability that the null hypothesis is true, and it does not measure the size or clinical importance of an effect. The registry’s p=0.103 is the sponsor-reported paired result. The unadjusted recalculation gives p=0.059 from posted summary statistics; it does not incorporate the participant-level pairing. The two values answer related but not identical statistical calculations.

For binary outcomes, an unadjusted comparison based on group-level proportions treats the observed counts as the available information and does not adjust for other variables. In this crossover setting, the recalculation is a simple summary-statistic comparison rather than a substitute for the paired trial analysis. It is useful as a check on the direction and scale implied by the posted marginal counts, but it cannot recover information omitted from those counts. The record does not report a fuller model or adjustment strategy.

Other measures in the registry include different sensor or plasma glucose cutoffs, whether rescue was administered, and glucose time in range. Binary participant counts answer whether an event occurred, while time-in-range outcomes describe a proportion of observation time. The registry lists these outcomes but the detailed results available for interpretation are not complete for every listed measure. Different thresholds also define different events; the result for one cutoff should not be treated as if it were a result for another.

Multiplicity arises when many outcomes or comparisons are examined. Even when each test is conducted at a conventional significance level, examining many results increases the chance of at least one apparently unusual result by chance. The registry record does not report a multiplicity adjustment or a hierarchy specifying which secondary outcomes should be treated as confirmatory. Thus, secondary p-values should be interpreted as unadjusted evidence unless the record states otherwise. The original statistical plan and any planned interim analyses are not reported in the registry record.

The primary analysis population is stated as the 12 participants who completed both study visits. The record does not describe an intention-to-treat analysis, a method for imputation, or sensitivity analyses for missing data. These omissions matter because the posted analysis is restricted to completers, and the reasons for non-completion are not given in the registry record reported. The sponsor’s paired analysis is the analysis reported by the trial; the unadjusted recalculation is an additional descriptive calculation from the posted totals, not a replacement for that analysis.

11. Confidence Intervals and Precision

The primary risk difference is -0.333, with a 95% confidence interval from -0.653 to -0.013. This interval conveys uncertainty about the unadjusted difference between the glucagon and control phases. It includes values near zero as well as more negative values, so the point estimate should not be read as a precise estimate of the size of the difference. The interval’s upper endpoint is close to zero, while its lower endpoint is much farther from zero.

The relatively small analysis population is important when considering precision. Only 12 participants completed both visits and contributed to the primary analysis, and the outcome counts were 1 and 5. With a limited number of participants, the observed difference can be sensitive to individual outcomes. The confidence interval makes that uncertainty visible, but it does not account for every limitation, including the absence of the participant-level paired pattern in the posted summary.

The reported paired McNemar result, p=0.103, does not come with a posted confidence interval in the registry record. The unadjusted interval and the paired p-value arise from different approaches to the data, so they should not be combined as if they were a single internally matched analysis. The record does not report a confidence interval for the paired treatment contrast.

Precision is not the same as importance. A narrow interval would indicate a more tightly estimated effect under the analysis assumptions, while a wide interval leaves more uncertainty about its magnitude. Neither interval width nor a p-value establishes the relevance of the result beyond the study conditions. Here, the estimate is based on a small completer group and short-interval challenges; conclusions should remain limited to the population and endpoint actually analyzed.

12. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The planning figures for the primary binary outcome are participants per arm, not total events. They are illustrations based on the observed effect, rather than the trial’s original design calculation. The registry’s planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect to detect is set to the stated fraction of the observed effect. The original design assumptions for this trial are not reported in the registry record.

The 80% power column describes the number of participants per arm that a hypothetical future study would need, under those planning assumptions, to have an 80% chance of detecting the chosen effect if that effect were real. The 90% power column uses a higher target chance of detecting that effect, and therefore calls for more participants per arm under the same assumptions. For planning at 100% of the observed effect, the illustration gives 26 participants per arm for 80% power and 34 for 90% power. For planning at 75% of the observed effect, it gives 51 and 68 participants per arm, respectively. For planning at 50% of the observed effect, it gives 125 and 167 participants per arm, respectively.

The pattern reflects a general principle: smaller effects are harder to distinguish from random variation, so detecting them requires more participants. These estimates depend on treating the observed effect as a planning reference and on the stated assumptions; an effect estimated from a small study can be unstable. They are not observed power for the completed trial. They are planning illustrations for a hypothetical study, not evidence about the probability that the completed study detected an effect.

13. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

n is the number of participants in each arm, p₀ and p₁ are the control and treatment proportions, and p̄ is their average.

Worked calculation for this trial

1. Control proportion p₀ = 0.417; treatment proportion p₁ = 0.083; average p̄ = 0.250.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.6124 = 1.2002.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.5652 = 0.4757.
4. n = (1.2002 + 0.4757)² / (-0.3333)² = 25.3, rounded up to 26 participants per arm for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
p0 <- 0.416666666666667                          # control proportion
p1 <- 0.0833333333333333                          # treatment proportion
n_per_arm <- function(power, frac = 1) {
  q1 <- p0 + (p1 - p0) * frac               # smaller effect = fraction of the observed difference
  pbar <- (p0 + q1) / 2
  z_b <- qnorm(power)
  ceiling(round((z_a * sqrt(2 * pbar * (1 - pbar)) + z_b * sqrt(p0 * (1 - p0) + q1 * (1 - q1)))^2 / (q1 - p0)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (unrounded n per arm)
cat("power.prop.test:", round(power.prop.test(p1 = p0, p2 = p1, power = 0.80)$n, 1), "\n")

# Simulation: how often does the z-test reject at the 80%-power sample size?
set.seed(3255629)
n    <- n_per_arm(0.80)
reps <- 20000
x0 <- rbinom(reps, n, p0); x1 <- rbinom(reps, n, p1)
ph <- (x0 + x1) / (2 * n)
z  <- ((x1 - x0) / n) / sqrt(ph * (1 - ph) * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(z) > z_a, na.rm = TRUE), reps))

Output from running the code

100% of observed effect: n per arm = 26 (80% power), 34 (90% power)
 75% of observed effect: n per arm = 51 (80% power), 68 (90% power)
 50% of observed effect: n per arm = 125 (80% power), 167 (90% power)
power.prop.test: 25.3 
Simulated power at n = 26 per arm: 0.848 (20000 replicates)

The simulation draws many pairs of event counts at the planned sample size and counts how often the two-proportion z-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

14. Analyses Reported by the Sponsor

The registry reports the sponsor’s primary analysis as a McNemar test comparing the glucagon and placebo/control phases, with p=0.103. McNemar’s test is suited to paired binary outcomes, which matches the crossover structure in which participants experienced both conditions. The posted analysis is therefore the trial’s own reported statistical result for the primary comparison.

The separate recalculation uses the posted marginal counts—1 of 12 in the glucagon phase and 5 of 12 in control—to derive an unadjusted risk difference of -0.333, a 95% confidence interval from -0.653 to -0.013, and p=0.059. Since the participant-level pairing is not available in the aggregate counts, this recalculation cannot reproduce the paired McNemar analysis. The distinct p-values should not be treated as contradictory estimates from the same method.

The registry record does not flag a discrepancy. The recalculation is best read as a descriptive check based on the summary counts, while the sponsor’s reported test is the analysis using the paired structure. The record does not provide the paired participant-level data or additional details that would allow a fuller reconciliation of the two calculations.

15. Multiplicity, Missing Data and Analysis Populations

The record includes a primary endpoint and a broad set of secondary outcomes, including event counts defined by multiple glucose thresholds, rescue administration, rebound hyperglycemia, and time in glucose ranges. When multiple outcomes are examined, the chance of finding an apparently unusual result by chance can increase. A prespecified hierarchy or statistical adjustment can help control how results are interpreted, but the registry record does not report whether either was used.

The original statistical analysis plan, including any multiplicity strategy, is not reported in the registry record. It is also not reported whether interim analyses were planned or conducted. In the absence of these details, the primary endpoint can be distinguished from secondary outcomes by its stated role in the registry, but the registry information does not establish which secondary analyses were confirmatory or how their error rates were managed. Unadjusted secondary results should therefore be treated as exploratory evidence.

The primary analysis included the 12 participants who completed both mixed-meal visits. Although 18 participants were enrolled, the record does not explain why other enrolled participants were not included in the primary analysis. It does not report how missing outcomes were handled, whether participants were analyzed according to randomized sequence, or whether an intention-to-treat population was used. No missing-data imputation or sensitivity analysis is described in the information reported.

These details matter because a completer analysis can differ from an analysis that includes all randomized participants, particularly if non-completion is related to outcomes or treatment experience. The registry record does not allow that possibility to be assessed. The paired structure also makes participant-level information important: the aggregate event counts do not show which participants had events in both phases or only one phase. Consequently, the sponsor’s paired analysis cannot be reconstructed from the posted margins alone.

16. 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.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: hypokalemia (5.6%).

The registry reports safety information for 18 participants in the randomized, double-blind, placebo-controlled crossover design. One participant was reported with a serious adverse event, and 16 participants were reported with other adverse events. No deaths were reported. The listed serious adverse event term was hypokalemia, reported as 5.6%.

These figures describe events recorded in the trial; they do not by themselves establish whether an event was caused by glucagon, placebo, the device, or another factor. The registry’s safety group is described across the treatment and control arms, and the posted figures do not assign each event to a particular phase in the information. The record also does not provide detailed event narratives, severity beyond the serious/non-serious classification, or a comparative causal analysis.

The reported counts should be read in the context of the small enrolled group and the study’s challenge-based crossover design. They are not a complete estimate of event frequency in broader use, and the absence of reported deaths does not resolve other safety questions. The registry record supplies the listed event counts and term but does not report additional details needed for a fuller assessment.

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

The posted primary counts show fewer participants meeting the sensor-glucose hypoglycemia definition in the glucagon phase than in the placebo/control phase among those included in the analysis. The unadjusted risk difference is -0.333, and its 95% confidence interval is -0.653 to -0.013. These figures describe the observed comparison under the study’s mixed-meal challenge conditions.

They do not establish that the intervention prevents hypoglycemia in every person with post-bariatric hypoglycemia, nor do they establish the size of an effect in settings outside the study challenges. The primary analysis includes 12 completers, and the record does not provide the reasons for the difference between enrollment and the analyzed group. The estimate is also based on participant counts at a specified sensor threshold, not on the full duration, severity, or daily pattern of glucose readings.

The sponsor’s paired analysis reports p=0.103, while an unadjusted recalculation from marginal counts reports p=0.059. Because these methods use different information about the paired crossover observations, they should not be read as interchangeable. The registry does not report enough participant-level detail to reconstruct the sponsor’s paired test from the summary counts.

The record does not establish long-term outcomes, effectiveness outside the study setting, or how the findings apply to people with characteristics not described in the posted trial information. Secondary outcomes provide additional measures but involve multiple analyses, and the registry does not report a multiplicity strategy. The results are therefore best understood as a small-study comparison of specified outcomes during controlled challenges, with uncertainty about both the magnitude and broader relevance of the observed differences.

18. Why This Trial Matters Statistically

This trial is statistically informative because its crossover structure makes the distinction between paired and unpaired analysis consequential. Each participant experienced both glucagon and placebo conditions, so the relationship between that participant’s two outcomes matters. Marginal event counts show how many participants had an event in each phase, but they do not show which participants changed status across phases. A paired method such as the reported McNemar test uses that within-participant structure.

The contrast between the sponsor-reported paired p-value and an unadjusted recalculation illustrates why summary counts are not enough to reproduce every analysis. The risk difference remains a useful description of the observed marginal contrast, but it does not substitute for the paired test. Reading both types of result with their methods clearly identified helps avoid treating a convenient recalculation as though it were the trial’s full analysis.

The trial also demonstrates why precision and analysis population deserve attention alongside a point estimate. The primary analysis includes 12 completers, and the interval around the unadjusted difference is broad. A result can have a visibly different count between phases while still leaving substantial uncertainty about the size of the underlying difference. The registry’s missing details about non-completion, missing-data handling, and the original analysis plan further constrain interpretation.

Finally, the range of secondary outcome definitions shows how conclusions can depend on exactly what is counted: sensor versus plasma glucose, different thresholds, rescue, or time in range. Such measures can provide complementary perspectives, but a collection of endpoints requires care about multiplicity and prespecification. The registry record does not report those details, making restrained interpretation important.

19. 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 primary analysis is based on 12 participants who completed both visits, compared with 18 enrolled. The registry record does not report why some enrolled participants were not included, how missing observations were handled, or whether sensitivity analyses were conducted. The resulting estimate may not describe all enrolled participants if completion was related to outcomes, but the available information does not allow that possibility to be evaluated.

The posted primary counts are marginal totals. They do not show the paired outcome pattern needed to reproduce the sponsor’s McNemar test, and no individual participant data are reported. The primary unadjusted risk difference and its interval therefore do not capture the full paired structure of the crossover design. The sponsor-reported paired p-value is available, but the registry does not report a confidence interval for that analysis.

The study evaluates specified glucose outcomes during mixed-meal challenges, with the second challenge within two weeks of the first. The primary endpoint is a binary sensor-glucose threshold and does not describe the duration or severity of low glucose readings. The registry summary mentions exercise, but the posted primary endpoint information concerns meal challenges. Results should not be extended to other settings or longer-term outcomes that are not reported.

The record does not report the original sample-size justification, detailed analysis plan, multiplicity procedures, or interim analyses. It lists numerous secondary outcomes, while complete detailed results are not available for every listed measure. These gaps limit assessment of which analyses were prespecified and how secondary findings should be weighted. Safety results are also presented as aggregate counts without detailed attribution by phase in the information reported.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Closed-Loop Glucagon Pump for Treatment of Post-Bariatric Hypoglycemia — NCT03255629.
• Registry record last updated 2022-09-06; status last verified 2022-08.
• Results first posted 2022-07-26.
• 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.

23. Record Summary

The registry record describes a completed randomized, quadruple-masked crossover study of a closed-loop glucagon pump for post-bariatric hypoglycemia. Participants were assigned to glucagon and matched placebo in randomized order during two mixed-meal challenges. Eighteen participants were enrolled; 12 completed both visits and were included in the primary analysis.

The primary endpoint counted participants with meal-provoked sensor glucose below <65 mg/dL. The posted counts were 1 of 12 during the glucagon phase and 5 of 12 during control. The unadjusted risk difference was -0.333, with a 95% confidence interval from -0.653 to -0.013. The sponsor reported a paired McNemar p-value of 0.103.

The registry also lists secondary glucose and rescue outcomes and aggregate adverse-event figures. The original sample-size rationale, missing-data procedures, multiplicity strategy, and fuller analysis plan are not reported in the registry record. Interpretation is limited by the small completer population, the distinction between paired and unadjusted analyses, and the challenge-specific outcome definitions.