The Effect of Continuous Glucose Monitoring on Glycemic Control in Insulin-requiring Gestational Diabetes Mellitus: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for The Effect of Continuous Glucose Monitoring on Glycemic Control in Insulin-requiring Gestational Diabetes Mellitus (NCT02204657), a randomized na study. The primary outcome is Glycemic Control by Measurement of HbA1c.

2. Clinical Question

3. Trial Design

This was a randomized, parallel-group trial with no masking, sponsored by the University of Malaya. The registry lists the study as completed and gives an enrollment of 50. The recorded start was 2013-04, and primary completion was 2015-04. The phase is listed as not applicable.

The two arms differed in glucose monitoring and in the stated basis for insulin titration. The continuous glucose monitoring arm received professional monitoring at 28, 32 and 36 weeks of gestation, with titration based on those results. The control arm did not receive continuous glucose monitoring and used fingerstick sugar readings to guide titration. The record does not describe whether participants, clinical staff, outcome assessors or analysts were masked; the overall masking designation is none.

Random allocation can help balance participant characteristics across groups on average, but the registry record does not report the sequence-generation method, allocation concealment, baseline characteristics or any adjustments used in analysis. It also does not report the trial's original sample-size assumptions or a detailed statistical analysis plan. These omissions limit how fully the design and analysis can be assessed from the record.

The primary outcome was mean HbA1c percentage from 28 weeks until delivery. Hypoglycemia, expressed as episodes per patient from 28 weeks until delivery, was a secondary outcome. The registry also records adverse-event counts. Although the study is marked completed, that status alone does not establish that every planned outcome or analysis was reported.

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 Glycemic Control by Measurement of HbA1c, measured over From 28 weeks until delivery.

HbA1c is a measure of glycemic control. It reflects glucose exposure over time rather than a single glucose reading, so it provides a summary measure that complements the glucose readings used to guide insulin titration. The registry defines the primary endpoint as the mean HbA1c percentage from 28 weeks until delivery.

A group mean gives a compact description of the average outcome in each arm. The spread reported with each mean is a standard deviation, which describes variation among participants; it is not the same thing as uncertainty in the estimated difference between group means. The comparison is reported as a mean difference, with continuous glucose monitoring contrasted against control.

HbA1c is relevant to the stated goal of glycemic control, but it does not directly measure hypoglycemia, the timing of glucose fluctuations, or every aspect of maternal and fetal health. The registry lists hypoglycemia separately as a secondary outcome. It does not provide a fuller definition of how HbA1c was collected or handled, including details about laboratory methods, repeated measurements or missing measurements. Those details are not reported in the registry record.

7. Results Posted in the Registry

• Continuous Glucose Monitoring System: mean 5.17 versus 5.57 on control; difference -0.40 (95% CI -0.68 to -0.12), standardized effect d = -0.81; z-test p = 0.004.

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.

Hypoglycemia

• Endpoint type: Event rate: events per unit of follow-up, compared as a rate ratio.
• Test: Poisson rate ratio, based on the estimated number of events and follow-up time.
• Result:
  • Continuous Glucose Monitoring System: 8.00 versus 4.00 (episodes per patient) on control; rate ratio 2.00 (95% CI 1.44–2.78, approximate, Poisson model); p ≈ <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference (a ratio of 1), so the event rate is statistically higher for Continuous Glucose Monitoring System than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Hypoglycemia is the only secondary outcome posted in the registry record. It addresses a distinct aspect of glycemic management from the primary HbA1c outcome. The overall pattern is mixed: the continuous-monitoring arm has the lower mean HbA1c, while its posted hypoglycemia comparison indicates a higher episode rate than control. These findings need not move in the same direction because a summary of average glycemic control does not capture the frequency of low-glucose episodes.

The secondary finding should be interpreted as supportive or exploratory evidence rather than as a stand-alone verdict about the intervention. The record identifies no multiplicity procedure or hierarchy for interpreting multiple outcomes, and no reported analyses are listed. When several endpoints are considered, the chance of finding an apparently notable result somewhere among them can increase. An unadjusted p-value does not account for that issue, so the secondary result carries less weight as confirmatory evidence than a clearly prespecified primary endpoint.

The registry's secondary measure is a median number of episodes per patient, with an interquartile range, while its comparison is a rate ratio. These are related but not identical summaries. The median describes a typical observed count; the rate ratio expresses a relative comparison under the stated count-model calculation. The record does not explain exposure-time handling or report checks for model assumptions, so the rate comparison warrants cautious interpretation.

Only one secondary endpoint is posted, which restricts what can be concluded about other potential consequences of monitoring and titration. The registry does not report additional secondary analyses or results for outcomes beyond hypoglycemia. The primary and secondary findings should therefore be considered together, while recognizing that they answer different questions and have different evidential status.

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

• Hypoglycemia

9. Statistical Interpretation

10. Statistical Methods Explained

What does the mean difference represent?

The primary outcome is continuous, and its group summaries are means with standard deviations. A mean is the arithmetic average of the measured values in a group. The mean difference expresses the contrast between the two group averages in the same units as the outcome: percentage points of HbA1c. The comparison is oriented as continuous glucose monitoring relative to control. A negative difference therefore corresponds to a lower mean in the monitoring arm in this record.

The standard deviations describe the spread of individual measurements around each arm's mean. They help characterize variation in the observed data, but they are not confidence intervals and should not be interpreted as the uncertainty around the estimated mean difference. The uncertainty in a group comparison depends on both the within-group variation and the amount of information in each group.

What does a confidence interval add?

A confidence interval gives a range of values compatible with the data under the method used to construct it. It is a way to assess the precision of an estimate, not a guarantee that the true effect falls within the particular interval. For the primary comparison, the reported interval is entirely on the negative side of zero. This indicates that the interval does not include no mean difference under the stated calculation. The interval remains an estimate based on the available summary data and does not account for unreported design or analysis issues.

The unadjusted p-value addresses how unusual a result at least as incompatible with a no-difference assumption would be under that assumption and the calculation's conditions. It is not the probability that the no-difference assumption is true, and it does not quantify the size or clinical importance of the difference. Because the record reports no analyses, the listed p-value should be treated as an unadjusted comparison rather than evidence of a documented, prespecified analysis plan.

Why are the secondary data different in form?

Hypoglycemia is summarized by a median and interquartile range rather than a mean and standard deviation. The median is the middle of the observed distribution, while the interquartile range describes its central spread. These summaries can be useful when episode counts are asymmetric or include a long tail. A median is not interchangeable with a mean: it describes the center of the distribution differently.

The secondary comparison is expressed as a rate ratio, and the posted calculation is marked as Poisson. In general, a Poisson approach models event counts in relation to exposure and assumes a particular relationship between the mean count and its variability. Over-dispersion occurs when event counts vary more than that model expects; if unaccounted for, uncertainty can be understated. The registry does not report a diagnostic assessment of this assumption, an alternative model or a sponsor analysis. The posted comparison should therefore be read with the limitations of its stated model in mind.

What does unadjusted mean?

An unadjusted comparison contrasts the groups without accounting for other variables in the analysis. In general, adjustment can address measured differences or improve precision when its use is appropriate and prespecified. It cannot automatically remove bias from unmeasured factors. The registry does not report adjusted analyses or the covariates that might have been considered. The results therefore do not establish whether an adjusted estimate would be similar.

How should recalculations be viewed?

Simple recalculations from posted summary statistics can help make an effect estimate, interval or planning illustration easier to interpret. Such calculations are not a substitute for the trial's own participant-level analysis. They may not reproduce methods that use repeated measures, covariate adjustment, clustering, missing-data procedures or other details unavailable in group summaries. The registry record lists no reported analyses, so the summary comparisons should be understood as limited to the available posted data and calculation methods.

Why do different outcomes need different summaries?

HbA1c is a continuous measurement, so differences in means are natural to interpret in the measurement's units. Hypoglycemia is an episode count, so its distribution and exposure period matter; the record uses medians and a rate-ratio comparison. Safety outcomes are reported as counts of participants affected. Each format answers a different statistical question, and none should be substituted for another. A favorable-looking result on one outcome does not mathematically resolve the findings for another outcome.

11. Confidence Intervals and Precision

The primary mean difference is -0.4, and its reported confidence interval lies entirely below zero. This interval communicates the uncertainty around the estimated difference using the calculation recorded for the comparison. Its position indicates that the interval is not compatible with a zero difference under that calculation, while its width still allows a range of possible effect sizes. The unadjusted p-value is 0.004, but the p-value does not replace the interval as a description of magnitude and precision.

The secondary rate-ratio estimate is 2.0, with a confidence interval from 1.441 to 2.776. This interval is above the no-ratio-difference value of one. The estimate and interval concern the rate comparison, not the difference between medians; those are distinct summaries. The posted p-value is <0.001 and is unadjusted.

Confidence intervals are useful because they show a range rather than a single point estimate, but their interpretation depends on the method and assumptions used. The registry does not report a full analysis plan, the handling of missing observations or diagnostic checks for the count model. Consequently, the intervals cannot address every source of uncertainty. They quantify sampling uncertainty under their calculation, not uncertainty from possible bias, incomplete outcome reporting or differences in care beyond the stated intervention.

The reported group sizes are 25 in each arm. The record does not report a separate precision target or a justification for the original enrollment. The interval widths should therefore be interpreted as properties of the posted results, not as evidence that the trial was designed to achieve a particular degree of precision.

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 are participants per arm for both the HbA1c comparison and the hypoglycemia comparison. These are not total event counts: neither comparison is a time-to-event calculation. The listed illustrations show planning for 100%, 75% and 50% of the observed effect, with columns for 80% and 90% power. For example, the primary-outcome illustration lists 25 participants per arm for 80% power when planning for 100% of the observed effect, and 43 participants per arm when planning for 75% of it. The secondary-outcome illustration lists 12 participants per arm for 80% power at 100% of the observed effect, and 22 at 75%.

The assumptions for these illustrations are a two-sided significance level of 0.05, equal allocation between the two arms and a normal approximation. The effect being targeted is the stated fraction of the observed effect. For the time-to-event case, the general planning convention would be total events, but it does not apply to these displayed comparisons. The original trial's own design assumptions and sample-size justification are not reported in the registry record.

Power is the probability that a planned analysis would detect an effect of the specified size under its assumptions if that effect were present. The 80% and 90% columns describe different target probabilities, not certainty that a future study will detect an effect. Smaller target effects require more participants under the stated assumptions, which is why the planning figures rise as the assumed fraction of the observed effect falls. These calculations are planning illustrations based on the observed results, not observed power.

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 = -0.400; pooled standard deviation σ = 0.496.
2. n = 2 × (1.96 + 0.842)² × 0.496² / -0.400² = 24.2, rounded up to 25 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 <- -0.4                      # observed difference in means
sigma <- 0.496487663492256                     # 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(2204657)
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 = 25 (80% power), 33 (90% power)
 75% of observed effect: n per arm = 43 (80% power), 58 (90% power)
 50% of observed effect: n per arm = 97 (80% power), 130 (90% power)
power.t.test: 25.2 
Simulated power at n = 25 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. Multiplicity, Missing Data and Analysis Populations

The registry names HbA1c as the primary endpoint and hypoglycemia as a secondary outcome. It does not report a multiplicity-adjustment procedure, a hierarchy for testing outcomes, or a broader statistical analysis plan. Without those details, the unadjusted secondary comparison should not be treated as if it had been protected against the increased chance of false-positive findings that can arise when multiple outcomes are examined. The record lists no reported analyses.

Outcome values are posted for both arms, but the registry does not describe participant flow, withdrawals, exclusions from analysis or the completeness of outcome collection. It does not report how missing measurements were handled, whether any imputation was used, or whether analyses followed intention-to-treat principles. The population used for each reported outcome is not described beyond the arm summaries and group sizes. These matters are not reported in the registry record.

Randomization is a design feature, but it does not by itself reveal how the analysis handled participants after allocation. If participants with unavailable measurements were omitted, the comparison could be affected if missingness differed between arms or related to outcomes. The record does not provide enough information to assess that possibility. Similarly, baseline balance, stratification, repeated-measure handling and covariate adjustment are not reported.

The registry does not report interim analyses, stopping rules, the original sample-size justification or a prespecified method for the primary comparison. Those omissions do not show that such procedures were absent; they mean that they are not reported in the registry record. As a result, the posted estimates can be interpreted descriptively, but the degree to which they reflect a fully prespecified analysis cannot be established from this record.

15. 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 zero participants affected by serious adverse events in each arm, with 25 at risk in each group. It also reports zero participants affected by other adverse events in each arm, again with 25 at risk in each group. These are counts of participants recorded as affected, not a complete description of every event type, severity or duration. No serious adverse-event terms are listed.

These figures should be read as no affected participants recorded in the reported categories during the available safety reporting, not as proof that the approaches carry no risk. The registry does not provide further detail on how adverse events were collected, the observation period for safety, or whether events were independently adjudicated. Death counts are not reported in the registry record; a missing death count should not be interpreted as a reported zero.

With a small trial and limited safety detail, absence of recorded events cannot establish equivalence in safety between arms. Safety counts are also separate from the primary and secondary efficacy summaries and should not be used to infer effects on outcomes that were not reported.

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

The posted primary result shows a lower mean HbA1c in the continuous glucose monitoring arm than in the control arm during the period from 28 weeks until delivery. The unadjusted estimated difference is -0.4, and the reported interval lies below zero. This is evidence of a difference in the recorded primary measure under the calculation used; it is not a complete account of all clinically important outcomes.

The secondary outcome points to a different aspect of care, with the posted comparison indicating more hypoglycemia episodes in the monitoring arm. The two outcomes can coexist: HbA1c summarizes average glycemic control, whereas hypoglycemia episodes describe low-glucose events. The registry does not report enough information to determine how the observed pattern relates to individual participants, timing, severity or duration of episodes.

The trial was randomized, which supports a comparison between allocated groups, but important details needed to evaluate that comparison are not reported in the registry record. These include baseline characteristics, missing-data handling, analysis populations and the statistical plan. The record also does not report broader maternal or fetal outcomes. The findings therefore should not be extended to outcomes not measured or posted, or to populations beyond the stated insulin-requiring gestational diabetes mellitus context.

The data do not establish that one approach is universally preferable, nor do they establish that the approaches are equivalent for safety. They describe group-level outcomes in this completed trial and leave uncertainty about precision, generalizability and unreported endpoints. The statistical results are one component of evidence, not a substitute for a full clinical assessment.

17. Why This Trial Matters Statistically

This trial illustrates why clinical comparisons benefit from examining more than a single outcome. HbA1c summarizes average glycemic control, while hypoglycemia captures episodes that may not be reflected by the average. The posted pattern differs across these measures, making it statistically important to distinguish the question each endpoint answers rather than treating all results as interchangeable evidence of benefit or harm.

It also demonstrates the value of reading effect estimates alongside uncertainty. A mean difference gives the direction and size of the observed contrast in the outcome's units, while a confidence interval conveys the range of estimates compatible with the calculation. A p-value alone cannot describe either the practical size of the difference or its precision. Similarly, count outcomes require attention to the choice of summary and model.

The record highlights the importance of transparent reporting. Randomization and a defined primary endpoint are useful design information, but assessing the credibility of an estimate also depends on details about allocation, outcome completeness, analysis populations, missingness and prespecification. When those details are not reported, the conclusions must remain narrower than the numerical results might initially suggest.

Finally, planning calculations based on observed effects can illustrate how target effect size affects required sample size, but they cannot recover the original design rationale or demonstrate that a study was adequately planned. Keeping planning illustrations separate from observed results avoids confusing a prospective design question with an after-the-fact description of trial data.

18. 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 gives limited information about eligibility criteria, participant characteristics and baseline comparability. Although the study is randomized, the method for generating and concealing the allocation is not reported. These details would help readers assess selection processes and the comparability of the groups.

Outcome reporting is also incomplete for evaluating the analysis. The record does not describe missing measurements, withdrawals, exclusions, analysis populations or handling of repeated observations. It does not report the original sample-size justification or a detailed statistical analysis plan. The posted analyses therefore cannot be checked against a prespecified protocol from the registry information alone.

The secondary hypoglycemia comparison is based on a count-model calculation marked as Poisson, but diagnostics and exposure-time details are not reported. The record also does not describe the severity or timing of episodes. For safety, the available counts have limited detail, and death counts are not reported in the registry record.

Generalizability is constrained by the record's description of the population as insulin-requiring gestational diabetes mellitus and by the trial's enrollment of 50. The registry does not report broader maternal or fetal outcomes, or follow-up beyond the stated measurement periods. The results should therefore be interpreted within the outcomes and population described, rather than extrapolated to other clinical settings or endpoints.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: The Effect of Continuous Glucose Monitoring on Glycemic Control in Insulin-requiring Gestational Diabetes Mellitus — NCT02204657.
• Registry record last updated 2016-08-05; status last verified 2016-06.
• Results first posted 2016-08-05.
• 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.

22. Record Summary

The registry describes a completed, randomized, parallel-group study sponsored by the University of Malaya, with no masking and an enrollment of 50. It compares professional continuous glucose monitoring with fingerstick-guided insulin titration in insulin-requiring gestational diabetes mellitus. The primary endpoint is mean HbA1c percentage from 28 weeks until delivery; hypoglycemia is the posted secondary outcome.

The continuous-monitoring arm had a lower reported mean HbA1c, with an unadjusted comparison whose confidence interval lies below zero. The secondary hypoglycemia comparison points in the opposite direction. Safety reporting lists no participants affected by serious or other adverse events in either arm, while death counts are not reported in the registry record.

Interpretation is limited by missing details on analysis plans, missing data, participant flow and outcome populations. The posted results describe a small randomized comparison, but do not resolve outcomes or design details that the registry does not report.