Reduction of Nocturnal Hypoglycemia and Hyperglycemia in the Home Using Predictive Algorithms: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Reduction of Nocturnal Hypoglycemia and Hyperglycemia in the Home Using Predictive Algorithms (NCT02438189), a randomized na study. The primary outcome is Percent Time Spent 70-180 mg/dL.

2. Clinical Question

The question concerns a measured glucose outcome during nights when the study systems were used. It is not a direct comparison of longer-term health outcomes. The registry’s brief summary describes the study’s aims as gaining experience with in-home use of the modified algorithm and gathering feasibility, safety, and initial efficacy data. Those aims set the context for interpreting the reported glucose summaries: the registry record documents a study of system use over a limited trial period, not evidence about every setting or every person with type 1 diabetes.

The registry lists 33 participants enrolled. The reported outcome summaries show 30 participants for each condition. Because nights were randomized within participants, the number attached to each condition should not be read as evidence that 30 separate people were assigned to each of two parallel groups. The registry record does not report the number of participants who contributed data to both conditions or the number of nights included in each participant’s outcome calculation.

3. Trial Design

This was a randomized, triple-masked, crossover study. The registry describes randomization at the night level within each participant: a participant could experience the study conditions on different nights, with the night being the unit of randomization. This differs from a parallel-group design, where participants are assigned to a single condition and remain in that group. The crossover approach makes within-participant comparisons possible in principle, but the registry record does not report the precise statistical model used to account for repeated nights or participant-level pairing.

The two conditions were predictive low glucose suspend (PLGS) and predictive hyper/hypo minimization (PHHM). PLGS used a predictive hypoglycemia minimization algorithm to suspend the pump. PHHM added a predictive hyperglycemia minimization algorithm that delivered automated correction boluses when estimated glucose was predicted to exceed 140 mg/dL, while also using predictive hypoglycemia minimization. The record labels the study triple-masked, but does not report which participants, investigators, or other roles were masked or how masking was maintained.

The registry lists enrollment of 33 and a planned sample size of 30 subjects in the brief summary. The primary and secondary outcome summaries report 30 participants for each condition. The reason for the difference between enrollment and the reported condition counts, and the participant flow leading to those summaries, are not reported in the registry record. The original sample-size justification and the trial’s design assumptions are also not reported.

The study is listed as phase NA and completed. It began in 2015-06 and had primary completion in 2016-03. The planned duration was approximately 3 months, including preliminary run-in activities and up to 90 days in the clinical-trial phase. The schedule included enrollment, assessments following continuous glucose monitoring and system run-in phases, the start of the clinical-trial phase, a 21-day visit, and a visit after 42 nights of successful system use. These planned milestones describe the study schedule; they do not establish that each participant completed the same number of nights or visits. The actual completion patterns are not reported in the registry record.

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: A run-in phase preceded the randomized trial. During the initial part of the run-in phase, the sensor was initiated and used for 14-21 days to verify that the participant could successfully use the pump and sensor. Successful participants then used the PHHM system at home for 5 nights to verify the ability to use it successfully.

Overall Study

Reasons for not completing this period:

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as Percent Time Spent 70-180 mg/dL, measured over Up to 42 nights.

The primary endpoint is the mean percentage of time that a participant’s glucose was between 70 and 180 mg/dL overnight, measured by continuous glucose monitor on each night. The registry also gives the equivalent range as 3.9–10.0 mmol/L and specifies a timeframe of up to 42 nights. In plain terms, the endpoint describes the share of monitored overnight time spent within the specified glucose interval, averaged as a continuous measure.

Time in range is used because it combines information about glucose values across monitored time into an interpretable summary. It reflects both the time spent within the chosen interval and, indirectly, time outside it; it does not identify whether out-of-range time was low or high. The registry separately lists measures of time spent in hypoglycemia and hyperglycemia among secondary outcomes, which can add information about those distinct directions.

The reported primary condition means are 71.0% for PLGS and 78.0% for PHHM, each with a standard deviation of 10.0. The unadjusted recalculated mean difference is 7.0 percentage points, with a 95% confidence interval from 1.939 to 12.06. The sponsor-reported analysis gives p < 0.001. These statistics describe the outcome as reported for this study; they do not by themselves establish how the conditions compare on outcomes that were not measured or whether the observed difference persists beyond the study period.

As a continuous endpoint, the measure can preserve gradations in overnight glucose exposure instead of reducing each night to a binary classification. Its meaning nevertheless depends on the specified glucose range, monitoring, and aggregation across nights. Details of data completeness, the handling of nights without usable monitoring data, and the exact participant-level averaging procedure are not reported in the registry record.

7. Results Posted in the Registry

• Predictive Hyper/Hypo Minimizer (PHHM): mean 78.00 versus 71.00 on control; difference 7.00 (95% CI 1.94 to 12.06), standardized effect d = 0.70; z-test p = 0.007.

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.

Mean Sensor Glucose

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard deviations.
• Result:
  • Predictive Hyper/Hypo Minimizer (PHHM): mean 143.00 versus 152.00 on control; difference -9.00 (95% CI -16.85 to -1.15), standardized effect d = -0.58; z-test p = 0.025.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Predictive Hyper/Hypo Minimizer (PHHM) than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Mean Home Glucose Meter Morning Glucose

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard deviations.
• Result:
  • Predictive Hyper/Hypo Minimizer (PHHM): mean 142.00 versus 163.00 on control; difference -21.00 (95% CI -31.45 to -10.55), standardized effect d = -1.02; z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Predictive Hyper/Hypo Minimizer (PHHM) than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Percent Time Spent 70 to 180 mg/dL (3.9 to 10.0 mmol/L) Overnight

• Endpoint type: Continuous: a mean value per participant, compared as a difference in means.
• Test: Unadjusted two-sample z-test for the difference in means, using the posted standard deviations.
• Result:
  • Predictive Hyper/Hypo Minimizer (PHHM): mean 78.00 versus 71.00 on control; difference 7.00 (95% CI 1.94 to 12.06), standardized effect d = 0.70; z-test p = 0.007.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically higher for Predictive Hyper/Hypo Minimizer (PHHM) than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

The secondary outcomes span average glucose, time in specified glucose ranges, and prolonged consecutive time in a low range. Among the secondary measures with group summaries, the direction is broadly consistent with the primary result: mean sensor glucose and morning meter glucose are lower under PHHM, while time in the target range is higher. The registry also lists 15 secondary outcomes, but summary values are not reported for every listed measure in the record information available for these comparisons.

These measures address related, but distinct, features of overnight glucose. An average glucose level does not describe how much time was spent within a target range, and a percentage of time below or above a threshold does not describe the duration of a continuous excursion. The primary outcome therefore cannot stand in for all secondary outcomes, and a pattern among the reported summaries does not establish a uniform result across every measure.

The unadjusted secondary comparisons carry less confirmatory weight than the primary result. Multiple outcomes create multiple opportunities for apparently notable findings, and the record does not report a multiplicity adjustment or a prespecified hierarchy for interpreting them. The registry reports its own two-sided test for mean sensor glucose, but does not list a sponsor analysis for morning meter glucose or the secondary time-in-range comparison. Recalculated p-values for these measures are not adjusted for the number of outcomes considered.

The secondary evidence is best treated as context for the primary endpoint rather than as a set of independent confirmations. The record does not report enough information about missing observations, participant-level outcome availability, or analysis populations to assess whether all secondary measures used the same data. No direction or magnitude should be imputed for secondary endpoints lacking reported group summaries.

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

• Mean Sensor Glucose
• Percent Time Spent <50 mg/dL (2.8 mmol/L)
• Percent of Time Spent <60 mg/dL (3.3 mmol/L)
• Percent Time Spent <70 mg/dL (3.9 mmol/L)
• Percent Time Spent >180 mg/dL (10.0 mmol/L)
• Percent Time Spent >250 mg/dL (13.9 mmol/L)
• Percent Time Spent >300 mg/dL (16.7 mmol/L)
• Percent Time Spent >30 Min and >60 Min Consecutive <50 mg/dL (2.8 mmol/L)

9. Statistical Interpretation

10. Statistical Methods Explained

The primary outcome is a continuous measure: the mean percentage of overnight time spent in the target glucose range. A mean gives a central summary of the observed values. The reported standard deviation describes how much the observed participant-level outcome values varied around their mean; it is not the uncertainty in the estimated mean. The primary means are 71.0% for PLGS and 78.0% for PHHM, with a standard deviation of 10.0 in each condition.

What does the mean difference express?

The unadjusted recalculation compares the PHHM mean with the PLGS mean. It reports a mean difference of 7.0 percentage points. A difference in percentages is expressed in percentage points, not as a relative percentage change. The estimated direction is positive because the reported mean is higher under PHHM. That description concerns the group summaries and does not mean every participant had a higher value in that condition.

What does the confidence interval add?

The recalculated 95% confidence interval runs from 1.939 to 12.06 percentage points. A confidence interval is a range of estimates compatible with the observed data and the calculation’s statistical assumptions. In repeated applications of the same method, intervals constructed this way would cover the target value at the stated rate. It is not correct to read an individual interval as assigning a probability to the fixed true value. The interval also does not show the spread of individual outcomes; the standard deviation addresses that different question.

Why does the crossover design matter?

In this study, nights were randomized within participants. Measurements from the same participant can be related: a participant’s outcomes under the two conditions may be more alike than outcomes from different participants. A paired or otherwise appropriately specified repeated-measures analysis can use that structure. An unadjusted recalculation from condition means and standard deviations does not include the within-participant covariance, the statistical quantity describing how paired outcomes vary together. It is therefore a summary-statistic calculation, not a substitute for a complete analysis of the night-level crossover data.

The registry reports a two-sided t-test for the primary comparison. A two-sided test evaluates departures in either direction rather than testing only for a higher or lower outcome. The registry’s reported result is p < 0.001. The unadjusted recalculated p-value is 0.007. Because the record does not provide the full analysis model, participant-level data, or a sponsor-reported estimate and confidence interval, the methods behind the registry result cannot be reconstructed from the summaries alone. The reported analysis and the simple recalculation answer related but not necessarily identical statistical questions.

How should the p-value be read?

A p-value measures how unusual data at least as incompatible with a specified null hypothesis would be if that null model and its assumptions were true. It is not the probability that the observed difference arose by chance, the probability that a treatment has no effect, or a measure of the size of the difference. It also does not indicate whether an effect matters clinically. The estimate and its uncertainty provide information about magnitude and precision that a p-value alone cannot provide.

What do the secondary comparisons show methodologically?

The secondary summaries use means and standard deviations for continuous measures. For mean sensor glucose, the reported means are 152.0 mg/dL under PLGS and 143.0 mg/dL under PHHM; the unadjusted recalculated difference is −9.0 mg/dL, with a 95% confidence interval from −16.85 to −1.152 and p=0.025. The registry reports p < 0.001 for its own two-sided t-test. For mean home glucose meter morning glucose, the reported means are 163.0 mg/dL and 142.0 mg/dL; the unadjusted recalculated difference is −21.0 mg/dL, with a 95% confidence interval from −31.45 to −10.55 and p < 0.001. No registry analysis is listed for that outcome. The primary time-in-range measure also appears among the secondary outcome summaries.

The secondary recalculations are unadjusted comparisons. A p-value from one of many secondary tests can be small even when none of the tests was selected as the sole confirmatory question. Testing multiple outcomes raises the chance of seeing apparently strong evidence somewhere by chance, unless the analysis accounts for multiplicity or the findings are interpreted as exploratory. The record does not report a multiplicity adjustment or a hierarchy for secondary hypotheses. Therefore, individual unadjusted secondary results warrant less confirmatory weight than a prespecified primary result, and they should be understood as supportive or exploratory evidence.

The registry lists outcomes involving mean sensor glucose, time spent at different glucose thresholds, and consecutive time in a low range. These endpoints are related but not identical: a mean level, the share of time in a range, and the duration of consecutive excursions capture different features of glucose monitoring. Results for one measure cannot be assumed to determine results for another. The available group summaries do not cover every listed secondary outcome, and missing values or analysis details are not reported in the registry record.

All statistical explanations above are general unless they describe a particular estimate from the registry. The summary-statistic calculations are explicitly unadjusted recalculations. They are useful for describing the reported group means and approximate uncertainty, but the crossover design and unavailable details limit how closely they can reproduce the sponsor’s analysis.

11. Confidence Intervals and Precision

For the primary comparison, the unadjusted recalculated mean difference is 7.0 percentage points, with a 95% confidence interval from 1.939 to 12.06. The interval gives a sense of the statistical precision of the estimated difference under the recalculation’s assumptions. Its width reflects uncertainty in the estimate; it does not say that participant outcomes lie within those limits. Individual variation is described separately by the reported standard deviations of 10.0 for both conditions.

A confidence interval is not a guarantee that the true difference lies inside its endpoints. In general, the stated confidence level describes the long-run performance of the interval-building procedure under its assumptions. The calculated interval is based on posted group summaries and does not include information about the pairing of nights within participants. Since this was a night-level crossover, a calculation that lacks the within-participant covariance may not reflect the precision of an analysis that uses the full crossover data.

The registry’s own primary analysis reports p < 0.001 but does not report an effect estimate or confidence interval for that analysis. The recalculated p-value is 0.007, paired with the interval above; it is not the sponsor’s p-value. The registry record does not report enough analysis detail to explain fully why the two p-values differ. The recalculated interval should therefore be read as the uncertainty from the simple summary-statistic comparison, not as an interval for the unreported sponsor model.

The secondary recalculated intervals also describe uncertainty under their own unadjusted calculations: the mean sensor glucose difference is −9.0 mg/dL, with an interval from −16.85 to −1.152; the morning meter glucose difference is −21.0 mg/dL, with an interval from −31.45 to −10.55. Those intervals do not account for multiple outcome testing, and their precision should not be confused with evidence that the results are independent of one another. The registry does not report intervals for its sponsor analyses of these outcomes.

Precision is only one part of interpretation. An interval can be relatively narrow while still being affected by limitations in the underlying design, measurement, or analysis. The registry record does not report a threshold for a clinically important difference, so statistical precision cannot determine practical importance on its own.

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 sample-size figures are planning illustrations based on the observed effect, not the original trial’s design calculation. For these continuous outcomes, figures are participants per arm. They are not total event counts. The 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 registry record does not report the assumptions used for the trial’s original sample-size justification; the brief summary states a sample size of 30 subjects.

The 80% power column means the planned chance of detecting an effect of the specified size under the planning assumptions, if that effect and the assumptions were true. The 90% power column represents a higher planned chance of detection for the same assumed effect, and generally requires more participants. These columns do not describe the probability that the observed result is true or the chance that the study’s conclusion is correct.

For the primary comparison, planning for 100% of the observed effect gives 33 participants per arm for 80% power and 43 participants per arm for 90% power. Planning for 75% of the observed effect gives 57 participants per arm for 80% power and 77 participants per arm for 90% power. Planning for 50% of the observed effect gives 129 participants per arm for 80% power and 172 participants per arm for 90% power. The pattern illustrates that smaller effects need more participants to detect under otherwise unchanged assumptions.

These are planning illustrations, not observed power. They use the effect estimated in this study as a reference, so they should not be taken as proof that a future study would observe the same effect or variability. The original design rationale, including its targeted effect and anticipated variation, is not reported in the registry record.

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, Δ is the difference in means to detect and σ is the common standard deviation.

Worked calculation for this trial

1. Observed difference Δ = 7.000; pooled standard deviation σ = 10.000.
2. n = 2 × (1.96 + 0.842)² × 10.000² / 7.000² = 32.0, rounded up to 33 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
delta <- 7                      # observed difference in means
sigma <- 10                     # pooled standard deviation
n_per_arm <- function(power, frac = 1) {
  ceiling(round(2 * (z_a + qnorm(power))^2 * sigma^2 / (delta * frac)^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 (t-based, so very slightly larger)
cat("power.t.test:", round(power.t.test(delta = delta, sd = sigma, power = 0.80)$n, 1), "\n")

# Simulation: exact two-sample t-test using simulated means and variances
set.seed(2438189)
n    <- n_per_arm(0.80)
reps <- 20000
m_diff <- rnorm(reps, delta, sigma * sqrt(2 / n))             # difference in sample means
s2     <- sigma^2 * rchisq(reps, 2 * (n - 1)) / (2 * (n - 1))   # pooled sample variance
tstat  <- m_diff / sqrt(s2 * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(tstat) > qt(0.975, 2 * (n - 1))), reps))

Output from running the code

100% of observed effect: n per arm = 33 (80% power), 43 (90% power)
 75% of observed effect: n per arm = 57 (80% power), 77 (90% power)
 50% of observed effect: n per arm = 129 (80% power), 172 (90% power)
power.t.test: 33 
Simulated power at n = 33 per arm: 0.797 (20000 replicates)

The simulation draws many sets of group means and variances at the planned sample size and counts how often the two-sample t-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 two-sided t-test with p < 0.001. It does not report the sponsor-analysis estimate or confidence interval. The unadjusted recalculation from the posted group summaries gives a mean difference of 7.0 percentage points, a 95% confidence interval from 1.939 to 12.06, and p=0.007. The recalculated values describe a direct comparison of the summary statistics, rather than a reconstruction of the full trial analysis.

The distinction matters because nights were randomized within participants in a crossover design. Participant-level pairing and repeated observations can affect the standard error and test statistic. The summary statistics do not include the within-participant covariance needed to reproduce a paired analysis. The registry names a two-sided t-test, but does not report enough detail about its implementation to establish how the night-level structure was handled or to reconcile the reported p-value with the simple recalculation.

The sponsor-reported result and recalculation should therefore be cited as separate results, not substituted for one another. The sponsor analysis is the registry’s reported inferential test; the recalculation is a transparent, unadjusted analysis of the available summaries. Neither p-value conveys the size of the result on its own, and the sponsor’s missing estimate and interval limit direct comparison of precision between the two analyses.

15. Multiplicity, Missing Data and Analysis Populations

The registry identifies a primary endpoint and lists secondary outcomes, but does not report a multiplicity strategy. It is therefore not known whether the sponsor adjusted secondary tests, used a hierarchical testing sequence, or treated the secondary analyses as exploratory. When many outcomes are tested, the chance of finding at least one apparently notable result can increase. This general concern is relevant to interpreting unadjusted secondary p-values, but the registry record does not establish which adjustment, if any, was part of the complete protocol.

Important details about missing data are also not reported in the registry record. It does not state how nights with incomplete continuous glucose monitoring were handled, what minimum amount of nighttime data was required, whether missing values were imputed, or how participants with incomplete condition exposure were included. The number of nights contributing to each participant’s average is not reported. As a result, the posted means cannot establish whether the amount or pattern of missing data differed between conditions.

The analysis population is not specified in enough detail to determine whether results use all enrolled participants, participants who completed system use, or another defined group. The record lists 33 enrolled and reports 30 participants per condition for the outcomes, but does not provide participant flow or explain the difference. It also does not report whether all participants contributed data to both crossover conditions.

The record states that nights were randomized within participants, but does not explain how the analysis accounted for repeated observations, participant-level pairing, study period, sequence, or possible carryover between conditions. The registry’s reported two-sided t-test does not provide enough detail to answer those questions. Interim analyses, stopping rules, and protocol deviations relevant to the statistical analysis are not reported in the registry record. These gaps limit assessment of how closely the reported analysis follows the design and how robust the estimates are to alternative assumptions.

16. 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 reports 30 participants at risk for serious adverse events in each condition. The number affected by a serious adverse event is reported as 0 for PLGS and 0 for PHHM. The registry likewise reports 30 at risk for other adverse events in each condition and 0 affected in each. No serious adverse-event terms are listed.

These are counts recorded in the registry for the reported conditions. They should not be interpreted as proof that no event of any kind occurred, nor as a comparison establishing that one condition is safer. The record does not report event-level details or explain how repeated exposure across crossover nights was handled in safety reporting. Death counts are not reported in the registry record.

Safety interpretation is also limited by the study context and duration: the registry describes an in-home trial with up to 42 nights of successful system use. The reported counts do not establish safety beyond the participants, conditions, and observation period represented in the record.

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

The registry results describe differences in measured overnight glucose outcomes between two study conditions among eligible participants. The primary outcome mean was 78.0% of time in the 70–180 mg/dL range under PHHM and 71.0% under PLGS. The unadjusted recalculation estimates a 7.0 percentage-point difference, with a 95% confidence interval from 1.939 to 12.06. The sponsor reports p < 0.001 for its primary two-sided test.

Those results do not, by themselves, establish that the difference is important to participants. The registry record does not report a threshold defining a clinically important change in time in range. A small p-value is not a measure of clinical importance, and the confidence interval reflects statistical uncertainty under the recalculation’s assumptions rather than all sources of uncertainty. The record also does not supply the sponsor analysis’s effect estimate or confidence interval.

The result is specific to the measured outcome, the two study conditions, and the people and nights represented in this trial. The trial was a night-level randomized crossover, with an intended period of up to 42 nights of successful system use. It does not establish effects on outcomes not reported in the registry, long-term outcomes, or outcomes in groups outside the eligibility criteria. The listed age range and other criteria restrict how directly the results can be generalized to people who do not meet them.

Nor do the summaries determine how the algorithms would perform in settings with different use patterns or support. The study required a companion to participate and be available at night. The registry’s stated aims included in-home experience, feasibility, safety, and initial efficacy; the glucose summaries address only part of that scope. Details of usability, adherence, and participant experience are not reported in the registry record.

The result should not be translated into a recommendation about treatment. It is a statistical description of posted study outcomes, with limits arising from the available reporting and the trial’s scope. Decisions about care require considerations beyond these registry summaries.

18. Why This Trial Matters Statistically

This study is statistically notable because it used night-level randomization within participants rather than assigning each participant to only one condition. A crossover structure can make within-person comparisons informative, since each participant experiences both conditions. It also makes the analysis more specialized: nights from the same participant are not necessarily independent, and the precision of a paired comparison depends on how closely outcomes under the conditions are related. The registry provides the basic design description but not enough analytical detail to assess fully how that dependence was handled.

The posted means show a higher primary time-in-range value under PHHM, and the recalculated confidence interval gives a way to assess uncertainty around a simple summary-statistic difference. At the same time, the sponsor’s reported p-value and the unadjusted recalculated p-value are not the same. That contrast illustrates why readers should distinguish the registry’s inferential test from calculations based solely on group summaries, especially when the design involves repeated observations.

The study also illustrates the difference between statistical evidence and statistical importance. P-values address compatibility with a null model under assumptions, while estimates and intervals communicate magnitude and precision. Neither decides whether an observed difference is meaningful to participants. The record does not report a clinically important threshold or the full sponsor effect estimate and interval.

Finally, the planning illustrations show how sample-size requirements depend on the effect a future study aims to detect. Smaller target effects require more participants under the stated planning assumptions. These calculations are separate from the original design rationale, which is not reported, and do not turn the trial’s observed findings into a forecast of future results.

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 registry record leaves key details of the crossover analysis unspecified. It describes night-level randomization within participants but does not report the model used to account for repeated nights, within-person pairing, study period, sequence, or possible carryover. This limits interpretation of the standard errors and makes the unadjusted recalculation less directly comparable with the sponsor’s reported analysis.

Participant and observation flow are not detailed. Enrollment is listed as 33, while outcome summaries report 30 participants per condition. The record does not explain this difference, state how many participants contributed to both conditions, or give the number of monitored nights contributing to each mean. Missing-data methods and the analysis population are not reported in the registry record.

The registry reports a sponsor p-value for the primary outcome but not the corresponding sponsor estimate or confidence interval. It also does not report the trial’s original sample-size justification, planned analysis assumptions, or a multiplicity strategy. Those omissions limit evaluation of the analysis plan and the degree to which secondary results were confirmatory.

The results concern a selected population: people with type 1 diabetes meeting the stated eligibility criteria, including the requirement for a companion who could participate and assist at night. The study was conducted in the home and its clinical-trial phase could last up to 90 days, with the primary outcome timeframe listed as up to 42 nights. The record does not establish how well the findings extend to people outside these criteria or to longer-term use.

Finally, the available secondary summaries cover only part of the listed outcome set. No data should be inferred for listed outcomes without reported group results. The registry’s safety counts are also limited to recorded affected participants and do not include death counts.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Reduction of Nocturnal Hypoglycemia and Hyperglycemia in the Home Using Predictive Algorithms — NCT02438189.
• Registry record last updated 2026-08-11; status last verified 2026-07.
• Results first posted 2026-08-11.
• 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

NCT02438189, titled “Reduction of Nocturnal Hypoglycemia and Hyperglycemia in the Home Using Predictive Algorithms,” is a completed randomized, triple-masked crossover study sponsored by the Jaeb Center for Health Research. The registry lists 33 participants enrolled, with randomization occurring at the night level within participants.

The study compared Predictive Low Glucose Suspend with Predictive Hyper/Hypo Minimizer. Its primary outcome was mean percentage of overnight time in the 70–180 mg/dL range, measured by continuous glucose monitoring for up to 42 nights. The reported means were 71.0% and 78.0%, respectively. The registry reports a primary two-sided t-test result of p < 0.001; an unadjusted recalculation from the group summaries gives a mean difference of 7.0 percentage points, with a 95% confidence interval from 1.939 to 12.06.

The registry also lists secondary glucose outcomes and reports zero participants affected by serious or other adverse events in each condition. Full details of the crossover analysis, missing-data handling, multiplicity plan, and original sample-size justification are not reported in the registry record.