A Research Study to Compare the Effect of Insulin Degludec and Insulin Glargine on Blood Sugar Levels in People With Type 2 Diabetes - the SWITCH PRO Study: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for A Research Study to Compare the Effect of Insulin Degludec and Insulin Glargine on Blood Sugar Levels in People With Type 2 Diabetes - the SWITCH PRO Study (NCT03687827), a randomized phase 4 study. The primary outcome is Percentage of Time Spent in Glycaemic Target Range 70-180 mg/dL (3.9-10.0 mmol/L) Both Inclusive, Using Flash Glucose Monitoring (FGM).

2. Clinical Question

3. Trial Design

This phase 4 study, NCT03687827, is titled the SWITCH PRO Study. The registry identifies it as randomized, open-label, multicentre and crossover. Participants with type 2 diabetes received insulin degludec and insulin glargine in separate periods; which treatment came first was decided by chance. Since both treatments were received by participants, the comparison is within a crossover structure rather than a simple comparison of two groups receiving only one treatment. The registry reports 498 participants enrolled.

The study started on 2018-10-02, reached primary completion on 2019-12-20 and has completed status. The brief summary describes a study lasting about 41 weeks, with 13 clinic visits and 27 phone calls. It also describes three periods of wearing a FreeStyle Libre Pro sensor, each for 2 weeks, and regular contact with study staff. The sensor readings were not visible to participants while they wore it.

The primary outcome uses flash glucose monitoring measurements during the maintenance periods in weeks 16-17 and weeks 34-35. The primary analysis population was per protocol, defined by remaining on assigned treatment and completing the assessments at visits 21 and 39, with the required minimum availability of measurements. The number analyzed for the primary outcome is listed as 448 for each treatment. In a crossover study, those treatment-specific counts should not be read as separate sets of 448 different participants.

Details of the randomization sequence, any washout period, the intended handling of period or carryover effects, and the original sample-size justification are not reported in the registry record. The trial's statistical plan beyond the reported analysis information is also 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: This was a cross-over trial. The trial included a 2-week run-in period (which was after the screening visit) for eligibility assessment in regard to adherence to FGM requirements. The participants were randomized into one of two treatment sequences.

Treatment Period 1

Reasons for not completing this period:

Treatment Period 2

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Per protocol (PP) analysis set - The PP analysis set consisted of subjects that stayed on assigned treatment until and completed FGM assessment at visits 21 and 39 respectively. Complete FGM assessment was specified as subjects having at least 70% of 2 weeks FGM measurements per maintenance period (2 \* 960 measurements).

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: Full analysis set (FAS) included all randomised participants.

6. Primary Endpoint

The registry lists the primary outcome as Percentage of Time Spent in Glycaemic Target Range 70-180 mg/dL (3.9-10.0 mmol/L) Both Inclusive, Using Flash Glucose Monitoring (FGM), measured over During the 2-week maintenance period: weeks 16-17 (period-1) and weeks 34-35 (period-2).

The primary endpoint is the percentage of time spent in the glycaemic target range of 70-180 mg/dL (3.9-10.0 mmol/L), with both limits included. It is calculated from flash glucose monitoring as the number of recorded measurements in that range divided by the total number of recorded measurements. The measurement window covers the two-week maintenance periods in weeks 16-17 and weeks 34-35.

This endpoint describes how much of the monitored time participants' recorded glucose measurements fell within a specified range. It is a continuous outcome expressed as a percentage of time, rather than a count of events or a single glucose reading. Monitoring over time offers a different summary from a blood test taken at one moment: it characterizes the distribution of recorded values across the monitoring period. The registry also specifies that at least 70% of planned flash glucose monitoring measurements had to be available during the maintenance periods for endpoint data to be included in the analysis.

General: A target-range measure can summarize glucose patterns across repeated measurements, but its interpretation depends on the selected range, monitoring coverage and analysis population. It does not by itself describe every aspect of glucose control or establish what caused a difference between treatment periods. The endpoint's numerical values are least-squares means, with standard errors reported as the measure of dispersion.

7. Results Posted in the Registry

• Insulin Glargine 100U/mL: mean 70.68 versus 72.11 on control; difference -1.43 (95% CI -3.48 to 0.62), standardized effect d = -0.09; z-test p = 0.172.

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.

Time Spent in Tight Glycaemic Target Range 70-140 mg/dL (3.9-7.8 mmol/L) Both Inclusive, Using Flash Glucose Monitoring

• 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 errors.
• Result:
  • Insulin Glargine 100U/mL: mean 51.45 versus 52.97 on control; difference -1.52 (95% CI -3.79 to 0.75), standardized effect d = -0.09; z-test p = 0.190.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Insulin Glargine 100U/mL and the control. The interval gives the range of effects that remain compatible with the data.

Time Spent in Nocturnal Glycaemic Target Range 70-140 mg/dL (3.9-7.8 mmol/L) Both Inclusive, in the Nocturnal Period (00:01 am - 05:59 am Both Inclusive) Using Flash Glucose Monitoring

• 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 errors.
• Result:
  • Insulin Glargine 100U/mL: mean 14.91 versus 15.15 on control; difference -0.24 (95% CI -0.96 to 0.48), standardized effect d = -0.04; z-test p = 0.514.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Insulin Glargine 100U/mL and the control. The interval gives the range of effects that remain compatible with the data.

Level of Glycated Haemoglobin (HbA1c) - Percentage

• 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 errors.
• Result:
  • Insulin Glargine 100U/mL: mean 7.16 versus 7.10 on control; difference 0.06 (95% CI -0.05 to 0.17), standardized effect d = 0.07; z-test p = 0.289.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Insulin Glargine 100U/mL and the control. The interval gives the range of effects that remain compatible with the data.

The secondary outcomes broadly address related aspects of glycaemic measurements: time in a narrower target range, time in the nocturnal portion of that range, HbA1c and mean glucose levels. The posted summaries for the continuous secondary outcomes show relatively similar treatment-period means in their reported units, and the displayed simple comparison intervals include zero for the outcomes with recalculated comparisons. This overall pattern does not provide a separate, consistent basis for stronger conclusions than the primary analysis.

These endpoints complement the primary measure rather than duplicate it. A narrower target range asks about a more restricted portion of the measurements; the nocturnal measure focuses on a specified time window; HbA1c is a distinct summary measure. Agreement or difference across such outcomes can help describe the pattern of results, but each outcome captures a different feature of the data. They should not be collapsed into one conclusion.

The registry reports no sponsor analyses for the posted secondary continuous comparisons. The recalculated comparisons are unadjusted and are based on summary statistics, so they do not replace trial-specific model results. The registry record also does not report a multiplicity adjustment or a full testing hierarchy. General: when several secondary outcomes are examined, nominal findings are more susceptible to chance findings unless a suitable adjustment or prespecified hierarchy is used. Accordingly, unadjusted secondary comparisons carry less confirmatory weight than a clearly specified primary analysis.

The record lists six outcome measures overall and five secondary outcomes posted. The available secondary results do not establish what would have happened on unreported or unposted outcomes, and they do not resolve the discrepancy between the sponsor's primary result and the simple primary recalculation. Their role is descriptive and supportive, interpreted with the outcome definitions, time windows and analysis methods in mind.

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

• Time Spent in Tight Glycaemic Target Range 70-140 mg/dL (3.9-7.8 mmol/L) Both Inclusive, Using Flash Glucose Monitoring
• Time Spent in Nocturnal Glycaemic Target Range 70-140 mg/dL (3.9-7.8 mmol/L) Both Inclusive, in the Nocturnal Period (00:01 am - 05:59 am Both Inclusive) Using Flash Glucose Monitoring
• Level of Glycated Haemoglobin (HbA1c) - Percentage
• Level of Glycated Haemoglobin (HbA1c) - mmol/Mol
• Mean Glucose Levels Using Flash Glucose Monitoring (FGM)

9. Statistical Interpretation

10. Statistical Methods Explained

General: A least-squares mean is a model-based mean estimate. In a trial with a crossover design, a statistical model may account for features such as treatment sequence or period, but the registry record does not provide enough detail to identify the model terms used for the sponsor's primary analysis. The posted least-squares means are 72.11 for insulin degludec and 70.68 for insulin glargine. The registry lists a standard error of 0.74 for each. A standard error describes the estimated uncertainty of a mean, rather than the spread of individual participants' measurements.

What does the simple recalculation represent?

The recalculated comparison is a simple, unadjusted comparison based on the posted treatment means and their summary information. It gives a mean difference of -1.43 when written as insulin glargine minus insulin degludec. Its 95% confidence interval is from -3.481 to 0.621, and its unadjusted p-value is 0.172. These are not a substitute for the sponsor's analysis. The registry explicitly indicates a discrepancy, and the sponsor's own analysis takes precedence. The recalculated comparison is useful as a transparent check of the posted summaries, but it may not reflect the model or contrast used in the trial's reported analysis.

The recalculated pooled standard deviation is marked as approximate. A standard deviation describes variability among observations, while the standard error describes uncertainty in an estimated mean. These are related but not equivalent quantities. The conversion used for a planning illustration depends on assumptions about the data and design. In particular, the crossover structure means that a simple independent-groups calculation is not a reproduction of a full crossover analysis. General: when outcomes vary more among participants, a given mean difference is harder to estimate precisely, all else equal; when repeated measurements within a participant are related, the analysis must account for that dependence to represent uncertainty correctly.

How should the reported test be read?

The registry lists a two-sided t-test for the sponsor-reported comparison. A two-sided test evaluates evidence for a difference in either direction, rather than only a prespecified direction. The reported p-value is 0.032, paired with a 95% confidence interval from 0.12 to 2.74 for the estimated difference of 1.43. Under the assumptions of that analysis, the p-value measures how unusual results at least as incompatible with the null hypothesis would be if the null hypothesis were true. It is not the probability that the null hypothesis is true, nor does it quantify the size or clinical importance of a difference.

General: a confidence interval gives a range of values compatible with the data and the statistical procedure, under its assumptions. It conveys uncertainty around the estimate, not a guarantee that the true value lies in the interval. A confidence interval that excludes zero for a difference corresponds to statistical evidence against a zero difference at the matching two-sided significance level, under the same model and assumptions. These concepts apply only to the analysis that generated the interval; they do not make differently constructed analyses equivalent.

What does over-dispersion mean?

General: over-dispersion refers to variation greater than a statistical model expects. It matters because a model that understates variability can produce standard errors that are too small and intervals that are too narrow. The primary endpoint is a continuous percentage derived from repeated sensor measurements. The registry does not report an over-dispersion assessment, a model diagnostic, or the particular variance structure used for the sponsor's primary analysis, so no claim about such checks can be made.

What can be inferred from the analysis population?

The registry defines the primary analysis set as per protocol. Participants had to stay on assigned treatment through the relevant assessment and meet the specified flash glucose monitoring coverage requirement. This conditions the analysis on treatment completion and adequate sensor data. General: restricting analysis to participants who meet such criteria can make the analyzed group differ from the full randomized population, especially if completion or measurement availability is related to treatment or outcome. The registry record does not report how missing outcomes were handled beyond the stated eligibility threshold, whether sensitivity analyses were conducted, or whether an intention-to-treat analysis was also reported.

How should multiplicity be considered?

The record lists a primary endpoint and multiple secondary outcomes, including other target-range measures and HbA1c. The registry does not report a multiplicity adjustment or a complete prespecified testing sequence. General: when multiple outcomes or comparisons are tested, the chance of obtaining at least one apparently notable result can increase if each test is treated independently at its nominal threshold. Without a stated adjustment or hierarchy, secondary p-values should not be treated as confirmatory evidence on the same footing as a single prespecified primary test. The trial's original statistical analysis plan and interim-analysis details are not reported in the registry record.

11. Confidence Intervals and Precision

The sponsor-reported primary estimate is 1.43, with a 95% confidence interval from 0.12 to 2.74. The interval lies on one side of zero and therefore indicates a positive estimated difference under the sponsor's contrast definition and analysis. Its width also shows that the data are compatible with a range of effect sizes, not only the point estimate. The registry does not provide enough detail to translate that range into a judgment about clinical importance.

The simple unadjusted recalculation has a different orientation and gives -1.43, with a confidence interval from -3.481 to 0.621. That interval includes zero. Because the registry marks a discrepancy, the sponsor's own analysis takes precedence; the recalculation should be read as a separate summary-statistic check rather than a competing authoritative result. The record does not explain the source of the difference between the analyses.

General: confidence intervals describe statistical precision under the method and assumptions used to calculate them. Narrower intervals indicate more precise estimates; wider intervals leave greater uncertainty about the magnitude. Precision is not the same as importance, and an interval does not account for biases or limitations that are not represented in the statistical model. The primary analysis also applies to the per-protocol population and the specified maintenance periods, not necessarily to all randomized participants or all times during the trial.

The posted standard errors for each treatment mean are 0.74. They are measures of uncertainty for those means, not confidence limits and not the variability of individual glucose readings. The reported treatment confidence interval is the more direct summary of uncertainty for the sponsor's contrast. The original sample-size justification and the trial's planned precision target are not reported in the registry record.

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 illustrations based on the observed effect, not a record of the trial's original design calculation. The listed assumptions are a two-sided significance level of 0.05, equal allocation between two arms, a normal approximation, and an effect to detect set at the stated fraction of the observed effect. The original trial sample-size assumptions are not reported in the registry record.

For these continuous outcomes, the planning quantities are participants per arm. They are not total participants across both arms. The 80% power column describes the number of participants per arm that the planning calculation estimates would be needed to have an 80% chance of detecting the specified effect, if the assumptions hold and that effect is present. The 90% column uses a higher chance of detection and therefore generally requires more participants per arm for the same effect.

The rows consider planning for 100%, 75% and 50% of the observed effect. A smaller effect is harder to distinguish from random variation, so the planning calculation requires more participants. For example, for the primary outcome, planning for 100% of the observed effect gives 1,884 participants per arm for 80% power and 2,522 participants per arm for 90% power; planning for 50% gives 7,533 and 10,085 participants per arm, respectively. These values are not evidence about what the completed trial was powered to detect under its original design.

General: power is a property of a planned statistical test under specified assumptions. It is not a measure of whether a completed study has proved an effect. These planning figures are not observed power; they are prospective illustrations using the posted effect estimate and stated calculation assumptions. The trial's crossover structure and sponsor analysis may not be represented by the simple planning model, so the figures should be interpreted as rough planning guidance rather than as a reconstruction of the trial protocol.

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 = -1.430; pooled standard deviation σ = 15.663.
2. n = 2 × (1.96 + 0.842)² × 15.663² / -1.430² = 1,883.2, rounded up to 1,884 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 <- -1.42999999999999                      # observed difference in means
sigma <- 15.6628477615024                     # 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(3687827)
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 = 1,884 (80% power), 2,522 (90% power)
 75% of observed effect: n per arm = 3,348 (80% power), 4,483 (90% power)
 50% of observed effect: n per arm = 7,533 (80% power), 10,085 (90% power)
power.t.test: 1884.2 
Simulated power at n = 1,884 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 comparison as an estimated treatment difference of 1.43, with a 95% confidence interval from 0.12 to 2.74 and a two-sided p-value of 0.032. The same estimated difference and confidence interval are also listed for a non-inferiority analysis, for which no p-value is shown. The result is labeled as a superiority analysis in one entry and a non-inferiority analysis in another.

The simple recalculation from the posted means uses the opposite contrast direction, insulin glargine minus insulin degludec, and gives -1.43. It also yields a confidence interval that crosses zero and an unadjusted p-value of 0.172. The registry marks a discrepancy between that calculation and the sponsor's reported result. The sponsor's own analysis takes precedence; the recalculation is not a correction to the registry result and should not replace it.

The record does not describe the sponsor model in enough detail to reproduce its analysis or account for the discrepancy. The non-inferiority margin, the full prespecified testing strategy and whether the superiority and non-inferiority analyses were planned in a particular order are not reported in the registry record. The reported estimates can therefore be quoted and described, but the underlying analysis cannot be independently reconstructed from the posted summary statistics alone.

15. Multiplicity, Missing Data and Analysis Populations

The registry lists a primary endpoint alongside several secondary outcomes. These include time in a tighter glycaemic range, time in that range during a nocturnal period, HbA1c in two units and mean glucose levels. The record does not report a multiplicity adjustment, a complete confirmatory testing hierarchy, or the protocol-defined order in which outcomes were to be tested. The registry record does not report interim analyses or the full statistical analysis plan. Without those details, the status of any secondary test as confirmatory cannot be established from the posted information.

General: multiplicity arises when more than one statistical hypothesis is tested. If several tests are each judged using the same nominal threshold without adjustment, the overall chance of at least one false-positive result can be greater than for a single test. Prespecified hierarchical testing or another adjustment can address this concern, but the chosen approach must be known to interpret results. An adjustment should not be assumed merely because the outcome list is available.

The primary outcome uses a per-protocol population requiring treatment completion and adequate flash glucose monitoring coverage during the maintenance periods. The record specifies a minimum of 70% of planned measurements and completion at visits 21 and 39. It does not report how observations from participants below that threshold were handled, whether missingness was imputed, whether sensitivity analyses were performed, or how missing values in other outcomes were treated. Those details are not reported in the registry record.

General: excluding participants who lack adequate measurements can affect interpretation if the likelihood of having usable data is related to treatment, health status or glucose values. A per-protocol analysis can also differ from an analysis retaining all randomized participants. These are reasons to examine analysis populations and missing-data methods, not grounds to assume that bias occurred. The registry record does not provide enough information to assess whether the reported analysis is robust to alternative missing-data assumptions or populations.

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 5% reporting threshold.

Most frequent serious adverse event terms

Serious events with the highest arm-level rate: pneumonia (0.6%), acute kidney injury (0.4%), acute coronary syndrome (0.2%), arteriovenous malformation (0.2%), atrial fibrillation (0.2%).

The registry lists serious adverse events among 490 participants at risk during insulin degludec exposure and among 484 participants at risk during insulin glargine exposure. It reports 26 participants affected in the degludec group and 23 in the glargine group. The record lists zero other adverse events in each treatment group in the posted summary and reports 0 deaths for degludec and 1 for glargine.

The listed serious-event terms include pneumonia at 0.6%, acute kidney injury at 0.4%, and acute coronary syndrome, arteriovenous malformation, atrial fibrillation and breast mass at 0.2% each. The registry summary does not provide enough detail in these listed terms alone to establish how events were distributed by treatment, how event terms were grouped, or whether an event was judged related to treatment.

General: serious adverse event counts describe events meeting seriousness criteria, not all symptoms or adverse experiences. They do not by themselves establish causation. In a crossover trial, participants may contribute safety exposure to both treatment periods, so treatment-specific counts at risk should not automatically be interpreted as numbers of unique participants randomized to separate groups. The registry record does not report the full safety analysis methods or a comparative statistical analysis of these counts.

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

The primary outcome concerns the share of monitored time with glucose readings in the specified target range during two maintenance periods. The posted treatment means are 72.11 for insulin degludec and 70.68 for insulin glargine. The sponsor reports an estimated difference of 1.43 with a 95% confidence interval from 0.12 to 2.74 and a two-sided p-value of 0.032. Those are the sponsor's stated results for the primary comparison. The simple unadjusted recalculation differs, and the registry marks that discrepancy; the sponsor's own analysis takes precedence.

These data describe a particular comparison: two insulin treatments in a randomized, open-label crossover study of people with type 2 diabetes, using flash glucose monitoring during the specified maintenance periods. The primary analysis was per protocol and required treatment completion and adequate sensor coverage. The interpretation therefore pertains to that analysis population and those periods, rather than automatically extending to every enrolled participant, other monitoring windows or other clinical outcomes.

A statistical difference does not by itself indicate whether the magnitude is important to patients or clinicians. The registry does not report a threshold for a clinically meaningful difference in percentage of time in range. The result also does not provide a complete picture of safety, long-term outcomes or every aspect of glucose control. Those questions require evidence beyond the primary endpoint alone.

General: a p-value is not the probability that a treatment is effective, and a confidence interval is not a guarantee about the true effect. Both are conditional on the analysis method and its assumptions. A registry result is a report of measured outcomes and analyses, not a substitute for the full protocol, statistical analysis plan or complete clinical context. The posted information supports description of the trial's outcomes but does not justify medical advice or a universal conclusion about treatment choice.

18. Why This Trial Matters Statistically

This trial is statistically notable for its crossover design and repeated glucose monitoring. Each participant received both treatments in different periods, allowing treatment-period outcomes to be compared within the trial structure. That design differs from a parallel trial in which participants remain in one assigned group, and analysis should account appropriately for treatment sequence and period. The registry identifies the crossover format but does not report the sponsor model details needed to assess how those features were handled.

The primary outcome is also a repeated-measure summary rather than a single laboratory measurement. Flash glucose monitoring estimates the proportion of recorded time within a defined range, while requiring sufficient sensor coverage for inclusion. This links the endpoint to both the outcome definition and the completeness of monitoring. The per-protocol analysis makes those eligibility details part of the interpretation.

The record highlights why reported model results and recalculations from summary statistics should not be conflated. The sponsor's analysis and the simple recalculation yield different inferential summaries, and the record flags a discrepancy. Stating that discrepancy openly, while giving precedence to the sponsor analysis, is important for a fair reading of the posted results.

Finally, the secondary outcomes cover related but distinct glycaemic measures, while the registry does not describe multiplicity handling or the full analysis plan. That combination illustrates a general issue in trial interpretation: a collection of outcome estimates is not equivalent to a fully specified inferential strategy. The record supports careful description of the primary result, its precision and population, while leaving important methodological details unresolved.

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.

Several trial-specific details needed for a full statistical assessment are not reported in the registry record. These include the randomization sequence procedure, the planned handling of period and carryover effects, any washout details, the sponsor's complete model specification, and the original sample-size justification. The complete statistical analysis plan and interim-analysis details are also not reported. Their absence limits independent reconstruction of the primary analysis.

The primary analysis is per protocol rather than an analysis explicitly described as including every randomized participant. Inclusion depended on remaining on assigned treatment, completing the specified assessments and meeting the sensor-data availability threshold. Missing-data handling beyond those criteria, sensitivity analyses and results for other analysis populations are not reported in the registry record. This makes it difficult to assess how the completion and monitoring requirements may affect the analyzed population.

The sponsor-reported primary comparison differs from the simple unadjusted recalculation, and the registry flags this discrepancy without explaining its source. The sponsor's own analysis takes precedence, but the lack of detailed model information prevents a full reproduction or reconciliation using the posted summaries alone. The non-inferiority margin is also not reported, limiting interpretation of the non-inferiority entry.

The record does not report a multiplicity adjustment or a complete testing hierarchy for the primary and secondary outcomes. Secondary comparisons are therefore best treated as descriptive unless further analysis-plan information is available. In addition, the trial is open-label, and the registry does not report whether outcome assessment or analysis was masked. The available information should not be used to infer that masking occurred. These limitations concern what can be assessed from the registry record; they do not establish that the trial's conduct or analyses were inadequate.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: A Research Study to Compare the Effect of Insulin Degludec and Insulin Glargine on Blood Sugar Levels in People With Type 2 Diabetes - the SWITCH PRO Study — NCT03687827.
• Registry record last updated 2021-01-12; status last verified 2020-12.
• Results first posted 2021-01-12.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

23. Record Summary

NCT03687827 is a completed phase 4 randomized, open-label, multicentre crossover trial sponsored by Novo Nordisk A/S. It enrolled 498 people with type 2 diabetes and compared insulin degludec 100U/mL with insulin glargine 100U/mL across treatment periods. The primary endpoint was percentage of time in the 70-180 mg/dL (3.9-10.0 mmol/L) range by flash glucose monitoring during maintenance periods in weeks 16-17 and weeks 34-35.

The primary per-protocol summaries list 448 participants for each treatment. The sponsor-reported estimated difference is 1.43, with a 95% confidence interval from 0.12 to 2.74 and p = 0.032. A simple unadjusted recalculation differs; the registry flags the discrepancy, and the sponsor's analysis takes precedence. The registry also posts secondary glycaemic outcomes and safety counts. Details of the original sample-size justification, full statistical analysis plan, multiplicity handling and several analysis methods are not reported in the registry record.