Safety and Efficacy Study of Albiglutide in Type 2 Diabetes: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Safety and Efficacy Study of Albiglutide in Type 2 Diabetes (NCT00849017), a randomized phase 3 study. The primary outcome is Change From Baseline (BL) in Glycosylated Hemoglobin (HbA1c) at Week 52.

2. Clinical Question

The registry describes a Phase 3, randomized, double-blind, parallel-group study. The primary outcome was a continuous measure, reported as a least-squares mean change in HbA1c. The sponsor’s reported comparisons use ANCOVA and give an estimate of the net mean difference for each dose versus placebo.

The central statistical question is about the size and precision of those between-group differences, rather than whether the observed changes alone establish a clinical benefit. The interpretation also depends on how the analysis handled baseline values, missing follow-up measurements, and values after hyperglycemic rescue. The registry record describes those details for the primary endpoint, while the original sample-size rationale and full multiplicity plan are not reported in the registry record.

3. Trial Design

This was a randomized, double-blind, placebo-controlled, parallel-group, multicenter Phase 3 study sponsored by GlaxoSmithKline. The registry lists the study as completed, with enrollment of 309. The recorded study start was 2009-01, and primary completion was 2011-11. Those dates describe the registry record and do not, by themselves, give the duration of participation for each individual.

The treatment groups represented in the primary endpoint results were placebo, Albiglutide 30 mg, and Albiglutide 50 mg. The intervention descriptions identify weekly injections and also describe albiglutide uptitration at Week 12. The registry record does not make clear which results arm corresponds to that uptitration description, so the arm descriptions should not be used to infer an unreported dose schedule.

Randomization is intended to make treatment groups comparable on average, while double masking is intended to limit the influence of treatment knowledge on participants and study personnel. A parallel-group design assigns participants to a group and compares outcomes across groups. These are general features of the design; the registry record does not report the randomization sequence method, allocation concealment details, or the specific procedures used to maintain masking.

The primary endpoint was assessed at Week 52. The registry lists additional outcomes at later time points, including HbA1c at Weeks 104 and 156, as well as outcomes concerning fasting plasma glucose, body weight, hyperglycemia rescue, and HbA1c response categories. The original sample-size justification, detailed protocol-level analysis plan, and any interim analyses 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: Participants (par.) who met eligibility criteria and completed a 4 week Run-in/Stabilization Period were then randomized to a 156-week Treatment Period, followed by 8 weeks of post-treatment follow-up. A total of 479 par. were screened; 309 par. were randomized, and 301 par. received \>=1 treatment dose.

Treatment Period (156 Weeks)

Reasons for not completing this period:

Follow-up Period (8 Weeks)

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Intent-to-Treat (ITT) Population with LOCF: all randomized par. who received \>=1 dose of study medication and who had a BL assessment and \>=1 post-BL assessment of HbA1c. Only par. with a value at BL and at the specified visit were analyzed. Values were carried forward for par. who were rescued or discontinued from active treatment before Week 52.

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 Change From Baseline (BL) in Glycosylated Hemoglobin (HbA1c) at Week 52, measured over Baseline and Week 52.

The primary endpoint was change from baseline in glycosylated hemoglobin, or HbA1c, at Week 52. The registry describes HbA1c as a form of hemoglobin used primarily to identify average plasma glucose concentration over a period of about 2 to 3 months. Baseline was the last non-missing value before treatment began, and change was defined as the Week 52 value minus the baseline value.

HbA1c is useful as a continuous endpoint because it captures a sustained aspect of glycemia rather than a single measurement at one moment. A comparison of mean change between randomized groups can therefore describe whether the groups followed different average trajectories over the study period. It does not show how every individual changed, and the average alone does not describe the full distribution of participant outcomes.

The registry reports least-squares means and standard errors for the treatment groups. Least-squares means are model-based averages adjusted for terms in the analysis; they are not simply the raw arithmetic averages. For this endpoint, the reported model included treatment group and specified baseline or participant factors, with baseline HbA1c included as a continuous covariate. The sponsor’s treatment contrasts are reported as net mean differences between each albiglutide dose and placebo.

The endpoint definition also specifies how missing post-baseline measurements and measurements after hyperglycemic rescue were handled. Those choices affect what the estimated Week 52 change represents and are relevant when interpreting the result.

7. Results Posted in the Registry

• Albiglutide 30 mg: mean -0.70 versus 0.15 on control; difference -0.85 (95% CI -1.12 to -0.58), standardized effect d = -0.89; z-test p = <0.001.
• Albiglutide 50 mg: mean -0.89 versus 0.15 on control; difference -1.04 (95% CI -1.31 to -0.77), standardized effect d = -1.09; z-test p = <0.001.

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.

Change From Baseline in Fasting Plasma Glucose (FPG) at Week 52

• 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:
  • Albiglutide 30 mg: mean -0.88 versus 1.00 on control; difference -1.88 (95% CI -2.54 to -1.22), standardized effect d = -0.79; z-test p = <0.001.
  • Albiglutide 50 mg: mean -1.38 versus 1.00 on control; difference -2.38 (95% CI -3.05 to -1.71), standardized effect d = -1.00; z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Albiglutide 30 mg than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Albiglutide 50 mg than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Change From Baseline in Fasting Plasma Glucose (FPG) at Week 156

• 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:
  • Albiglutide 30 mg: mean -1.31 versus -0.23 on control; difference -1.08 (95% CI -1.84 to -0.32), standardized effect d = -0.71; z-test p = 0.006.
  • Albiglutide 50 mg: mean -1.83 versus -0.23 on control; difference -1.60 (95% CI -2.41 to -0.79), standardized effect d = -0.92; z-test p = <0.001.
• Interpretation:
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Albiglutide 30 mg than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.
  • The 95% confidence interval excludes no difference, so the mean is statistically lower for Albiglutide 50 mg than for the control. This is an unadjusted, secondary comparison, so it is suggestive rather than confirmatory.

Change From Baseline in Body Weight at Week 52

• 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:
  • Albiglutide 30 mg: mean -0.39 versus -0.66 on control; difference 0.27 (95% CI -0.91 to 1.45), standardized effect d = 0.06; z-test p = 0.654.
  • Albiglutide 50 mg: mean -0.86 versus -0.66 on control; difference -0.20 (95% CI -1.39 to 0.99), standardized effect d = -0.05; z-test p = 0.742.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Albiglutide 30 mg and the control. The interval gives the range of effects that remain compatible with the data.
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Albiglutide 50 mg and the control. The interval gives the range of effects that remain compatible with the data.

The posted secondary outcomes cover additional measures of glycemic change, weight, time to hyperglycemia rescue, and the number of participants reaching specified HbA1c response thresholds. The outcomes span more than one follow-up time, including later assessments. This broadens the description of the trial beyond its Week 52 primary endpoint, but it also creates a wider set of comparisons to interpret.

In the posted numeric secondary results, fasting plasma glucose at Week 52 has lower reported mean changes in both albiglutide groups than in placebo, and the unadjusted comparisons are negative. The Week 156 fasting plasma glucose summaries also have more negative mean changes in the albiglutide groups than in placebo, but the reported group sizes are smaller than at Week 52. The Week 52 body-weight summaries do not show a consistent separation of both active arms from placebo, and the unadjusted comparison intervals include zero. These broad patterns are compatible with a more pronounced primary HbA1c contrast than the displayed Week 52 weight contrasts, but they are not a substitute for the primary endpoint analysis.

The secondary measures address related but distinct questions. Fasting plasma glucose is a different measure from HbA1c, body weight is not a glycemic endpoint, and reaching a response threshold is not the same analysis as comparing mean change. Secondary outcomes can add context, but should not be combined into a single claim about treatment effect.

Multiplicity matters because testing many outcomes, groups, and time points increases the chance of obtaining at least one apparently strong result by chance. The registry record does not report a multiplicity adjustment for these secondary analyses. Accordingly, unadjusted secondary results carry less inferential weight than a prespecified primary analysis with a defined model. The outcome list also includes measures for which numeric results are not described in the registry entries; their results are not reported in the registry record.

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

• Change From Baseline in HbA1c at Weeks 104 and 156
• Time to Hyperglycemia Rescue
• Change From Baseline in Fasting Plasma Glucose (FPG) at Week 52
• Change From Baseline in Fasting Plasma Glucose (FPG) at Week 156
• Number of Participants Who Achieved Clinically Meaningful HbA1c Response Levels of <6.5%, <7%, and <7.5% at Week 52
• Number of Participants Who Achieved Clinically Meaningful HbA1c Response Levels of <6.5%, <7%, and <7.5% at Week 156
• Change From Baseline in Body Weight at Week 52
• Change From Baseline in Body Weight at Week 156

9. Statistical Interpretation

10. Statistical Methods Explained

The primary endpoint analysis used analysis of covariance, commonly abbreviated ANCOVA. In general, ANCOVA compares outcome averages across groups while accounting for selected covariates. In this trial, the model included treatment group, region, history of prior myocardial infarction, and age category as factors, and baseline HbA1c as a continuous covariate. The age categories were described as younger than 65 years versus 65 years or older, and prior myocardial infarction was represented as yes versus no.

Adjusting for baseline HbA1c can improve the comparison by accounting for variation in starting values. The model’s least-squares means are adjusted averages conditional on the model terms, and the reported net mean difference is the estimated contrast between an albiglutide group and placebo. The sponsor’s reported estimates and confidence intervals come from this ANCOVA approach. A re-calculation from group summaries is not the same model: it is a simple, unadjusted comparison and cannot reproduce all adjustments made in ANCOVA.

The primary outcome is continuous, measured as change in HbA1c percentage from baseline to Week 52. The registry reports standard errors alongside the group least-squares means. A standard error describes the estimated uncertainty in a mean; it is not the same quantity as the standard deviation, which describes variation among individual observations. This distinction matters when reading the endpoint summaries and any uncertainty intervals based on them.

The registry specifies last observation carried forward, or LOCF, for missing post-baseline HbA1c values. The last non-missing post-baseline on-treatment measurement was used to fill in a missing measurement. HbA1c values obtained after hyperglycemic rescue were treated as missing and replaced with pre-rescue values. The stated primary analysis population was the Intent-to-Treat population with LOCF, described as randomized participants who received at least one dose, had a baseline assessment, and had at least one post-baseline assessment. The record also says that only participants with a baseline value and a value at the specified visit were analyzed.

LOCF is a missing-data method that carries a previous observed value forward when a later value is missing. As a general statistical matter, this approach can be problematic if the outcome would have changed after the last measurement; carrying a value forward does not incorporate uncertainty about what the unobserved value might have been. This general limitation does not establish the direction or size of any bias in this trial. The registry record does not report sensitivity analyses that would show how the primary result changes under other missing-data assumptions.

Hyperglycemic rescue measurements were handled separately: values after rescue were set aside for the primary endpoint and replaced with pre-rescue values according to the recorded method. This means the estimand represented by the analysis is not simply the observed HbA1c value regardless of rescue. The registry record does not provide a fuller estimand statement describing all treatment discontinuation or intercurrent-event strategies.

The reported p-values test a null hypothesis of no difference under the model’s assumptions. A small p-value indicates that data at least as incompatible with that null were uncommon under the specified test; it does not measure the probability that the null is true, the probability that the treatment effect is clinically important, or the probability of replication. Confidence intervals add an estimate of the range of values compatible with the data and model at the stated confidence level.

For general interpretation, over-dispersion means that observed variability is greater than a model expects. It can make standard errors too small if the model fails to account for that extra variability. The primary outcome is continuous, and the registry does not report an over-dispersion assessment or a specific correction for it; this concept should not be read as a claim that over-dispersion occurred in this trial.

The registry reports two sponsor ANCOVA comparisons for the primary endpoint, each against placebo. It does not report whether or how the sponsor adjusted for multiple comparisons across doses, outcomes, or time points. The original statistical analysis plan, interim-analysis plan, and original sample-size assumptions are not reported in the registry record. Those omissions limit what can be concluded about the broader inferential strategy beyond the reported endpoint model and results.

11. Confidence Intervals and Precision

A confidence interval expresses uncertainty about an estimated effect under the statistical model. It gives a range of values compatible with the data and model at the stated confidence level; it is not a range containing a fixed proportion of individual participants’ outcomes. A narrower interval indicates greater statistical precision, while a wider interval indicates more uncertainty about the size of the mean difference.

For the primary comparison of Albiglutide 30 mg with placebo, the sponsor reported a 95% confidence interval from -1.11 to -0.58. For Albiglutide 50 mg versus placebo, the reported 95% confidence interval was -1.31 to -0.77. Both intervals are below zero, and the sponsor reported p < 0.001 for each comparison. These intervals support a directionally negative fitted mean difference while still allowing a range of possible magnitudes.

The interval does not establish whether the estimated difference is large enough to matter clinically. The registry record does not report a threshold for clinical importance, so statistical precision should not be confused with clinical interpretation. Nor does the interval capture every source of uncertainty: it reflects the specified analysis and does not automatically account for unreported choices such as multiplicity procedures or alternative missing-data assumptions.

The sponsor’s reported intervals are based on ANCOVA. The simple re-calculation uses posted summary statistics and is not a replacement for the adjusted model. Differences between the two approaches can arise because they use different information and assumptions. In this case, the re-calculated point estimates are close to the sponsor’s reported point estimates, but the sponsor’s analysis remains the primary reported inference.

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 illustrations are stated as participants per arm because the outcomes in this analysis are continuous. They are not total event counts. Each row asks how many participants per arm would be needed under a planning calculation if the effect to detect were 100%, 75%, or 50% of the observed effect. The columns for 80% and 90% power describe the sample size associated with those chosen power targets: in general, higher target power requires a larger planned sample for the same effect and assumptions.

The listed planning assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The target effect is the stated fraction of the observed effect. The illustrations are based on the observed treatment contrast and are not the trial’s original design justification; that original justification is not reported in the registry record.

Smaller effects are harder to distinguish from random variation, so the planning calculations require more participants per arm when the target is a smaller fraction of the observed effect. The displayed curve should be read as a visual version of this same relationship, not as a record of the number of participants actually enrolled or analyzed. These values are not observed power; they are planning illustrations conditional on the assumptions and effect sizes used.

In general, a planning calculation helps show how sample size, effect size, and a chosen power target relate. It does not establish that a future study will obtain the assumed effect, nor does it account for unreported design choices or departures from the stated calculation assumptions.

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 = -0.850; pooled standard deviation σ = 0.960.
2. n = 2 × (1.96 + 0.842)² × 0.960² / -0.850² = 20.0, rounded up to 21 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 <- -0.85                      # observed difference in means
sigma <- 0.960124231970179                     # 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(849017)
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 = 21 (80% power), 27 (90% power)
 75% of observed effect: n per arm = 36 (80% power), 48 (90% power)
 50% of observed effect: n per arm = 81 (80% power), 108 (90% power)
power.t.test: 21 
Simulated power at n = 21 per arm: 0.804 (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 sponsor reported ANCOVA estimates of -0.84 for Albiglutide 30 mg versus placebo and -1.04 for Albiglutide 50 mg versus placebo. The corresponding 95% confidence intervals were -1.11 to -0.58 and -1.31 to -0.77, and both reported p-values were < 0.001.

The simple unadjusted re-calculations from posted summary statistics estimated differences of -0.85 and -1.04, respectively. These estimates are close to the sponsor’s reported values. The re-calculation is useful as a transparent check based on group summaries, but it does not reproduce the sponsor’s ANCOVA adjustment for treatment group, region, prior myocardial infarction history, age category, and baseline HbA1c.

The registry record does not flag a discrepancy between the sponsor result and the re-calculation. The sponsor’s model-based analysis is therefore the relevant primary analysis to interpret, while the recalculation offers a secondary check rather than an alternative trial result. The registry also does not report the full analysis code or a broader protocol-level statistical plan, so the reported model description sets the limit of what can be assessed about the sponsor’s analysis.

15. Multiplicity, Missing Data and Analysis Populations

The primary analysis population is described as an Intent-to-Treat population with LOCF, including randomized participants who received at least one dose, had a baseline HbA1c assessment, and had at least one post-baseline HbA1c assessment. The record further states that only participants with baseline and specified-visit values were analyzed, with carried-forward values used for participants who were rescued or had missing measurements under the described method. This definition makes clear that the analysis is not simply based on every randomized participant regardless of available endpoint information.

Missing post-baseline HbA1c values were handled using the last non-missing post-baseline on-treatment value. Values after hyperglycemic rescue were treated as missing and replaced with pre-rescue values. As a general statistical matter, LOCF assumes that the last observed value is an acceptable stand-in for the missing measurement. If the outcome would have changed after that observation, the carried-forward value may not reflect the unobserved outcome. The registry record does not report sensitivity analyses using alternative missing-data methods.

The record lists 13 outcome measures and 12 posted secondary outcomes. The primary endpoint includes two dose-versus-placebo comparisons, and the outcomes also cover multiple measures and follow-up times. The registry record does not report whether the sponsor adjusted for multiplicity, which outcomes were included in any formal testing hierarchy, or how the type I error rate was controlled across comparisons. Without that information, secondary p-values should not be read as if each were the sole confirmatory test.

The record does not report interim analyses, the original sample-size justification, or the complete protocol-level statistical plan. It also does not report enough information to assess all analysis-population choices for every secondary measure. These are reporting limitations; they do not demonstrate that any particular analysis choice was inappropriate.

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 2% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: cardiac failure congestive (2.0%), coronary artery disease (2.0%), acute myocardial infarction (2.0%), transient ischaemic attack (2.0%), chest pain (2.0%).

The registry reports participants affected by serious adverse events and by other adverse events in each safety group. For serious adverse events, the counts were 16 of 101 in placebo, 15 of 101 in Albiglutide 30 mg, and 14 of 99 in Albiglutide 50 mg. For other adverse events, the counts were 78 of 101, 84 of 101, and 83 of 99, respectively. These are participant counts as recorded, not a statistical comparison of risk.

The serious-event terms listed most often include cardiac failure congestive, coronary artery disease, acute myocardial infarction, transient ischaemic attack, and chest pain, each reported as 2.0%; breast cancer was reported as 1.0%. The record does not provide a formal comparative analysis of these terms in the registry record. The death counts are not reported in the registry record.

These figures are descriptive. They do not establish that an event was caused by a study treatment, and they should be interpreted alongside the study’s follow-up, event definitions, and complete safety reporting. A difference or similarity in raw counts alone is not a causal conclusion.

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

The primary endpoint results show that the sponsor’s ANCOVA estimated negative mean differences in Week 52 HbA1c change for both albiglutide dose groups compared with placebo. The confidence intervals reported for both comparisons lie below zero, and the sponsor reported p-values of < 0.001. This is evidence about the modeled average contrast under the endpoint definition and analysis methods in the registry record.

The result does not describe the response of every participant. Group means can coexist with substantial variation among individuals, and the posted least-squares means do not show the full distribution of outcomes. The result also does not determine whether the estimated difference crosses a threshold of clinical importance; such a threshold is not reported in the registry record.

The interpretation is specific to the Week 52 endpoint, its analysis population, and its treatment of missing and post-rescue values. It does not by itself establish what happened at other time points, how every secondary endpoint should be interpreted, or how the estimate would change under different missing-data assumptions. The registry’s secondary outcomes provide additional context but include multiple comparisons, and a multiplicity adjustment is not reported in the registry record.

The statistical findings should not be converted into medical advice or a general conclusion about treatment for individuals. The trial compares group-level outcomes in the enrolled study population under the recorded conditions. Its findings do not automatically apply to populations, treatment schedules, or settings not described in the registry record.

Finally, a small p-value is not a measure of the probability that the treatment effect is real, nor does it convey the effect’s clinical importance. The point estimate, interval, endpoint definition, missing-data approach, analysis population, and broader study context all contribute to a careful reading.

18. Why This Trial Matters Statistically

This trial illustrates why a clinical result should be read through its statistical structure rather than by focusing on a single p-value. The primary endpoint is a continuous measure, the reported group values are model-adjusted least-squares means, and the sponsor’s contrasts come from ANCOVA. That distinction matters because a simple calculation from group summaries and an adjusted model answer related but not identical questions.

It also illustrates the role of baseline adjustment. HbA1c at baseline is included as a continuous covariate, while other stated participant and site factors are included in the model. Such adjustment can account for measured differences and improve precision, but it does not remove the need to understand which population was analyzed and how missing values were handled.

Missing-data rules affect the meaning of an endpoint estimate. Here, LOCF and the handling of values after hyperglycemic rescue are specified. These details make the analysis more interpretable, while also identifying assumptions that a reader may want to examine through sensitivity analyses. The registry record does not report such analyses.

The study also provides a useful contrast between a primary endpoint and a collection of secondary measures. Secondary outcomes can broaden understanding, but several measures and time points create multiplicity concerns. When an adjustment or testing hierarchy is not reported, the primary analysis should carry more weight than isolated secondary findings.

Lastly, the planning illustrations demonstrate how power targets and assumed effect sizes inform sample-size calculations. They are distinct from the trial’s original design rationale and from observed power. Separating planning calculations from the trial’s actual analysis helps prevent an observed result from being mistaken for evidence that the study was adequately designed for every outcome.

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 provides a primary analysis description, but the original sample-size justification is not reported. The sample-size illustrations based on observed effects should therefore not be treated as evidence of the assumptions used to design the trial.

The primary endpoint used LOCF for missing post-baseline measurements and substituted pre-rescue values for measurements after hyperglycemic rescue. These methods are explicitly described, but the registry record does not report sensitivity analyses using alternative assumptions. The available information therefore does not show how robust the estimated effect is to other ways of handling missing values or rescue.

The record does not report a multiplicity strategy, a complete testing hierarchy, or interim analyses. This is particularly relevant because multiple outcomes, dose comparisons, and follow-up times are listed. It limits interpretation of secondary findings as confirmatory evidence.

The primary endpoint summaries include fewer participants in each arm than the total enrollment of 309, and the analysis-population description requires specified baseline and follow-up measurements. This means the endpoint-specific summaries should not be mistaken for a summary of every enrolled participant. The registry record does not provide enough detail in the registry record to fully assess missingness patterns or reasons for non-inclusion in each endpoint analysis.

There are also limits to what the safety figures establish. They are counts of participants affected, with no formal comparative safety analysis; death counts are not reported in the registry record. The registry’s intervention descriptions mention uptitration, but do not map that description to a specific results arm. These reporting gaps should be kept distinct from the statistical findings that are reported.

20. Related Tutorials

21. Related Statistical Calculators

22. Sources

• ClinicalTrials.gov: Safety and Efficacy Study of Albiglutide in Type 2 Diabetes — NCT00849017.
• Registry record last updated 2017-01-09; status last verified 2016-11.
• Results first posted 2014-06-05.
• 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.

23. Record Summary

The registry record describes a completed, randomized, double-blind, parallel-group Phase 3 study of albiglutide and placebo in type 2 diabetes, with enrollment of 309. The primary outcome was change from baseline in HbA1c at Week 52, analyzed using ANCOVA with specified factors and baseline HbA1c as a covariate.

The sponsor reported negative net mean differences for both albiglutide dose comparisons with placebo, with 95% confidence intervals below zero and p-values of < 0.001. The unadjusted re-calculations from posted summaries were close to the sponsor’s estimates. Secondary outcomes cover additional glycemic, weight, rescue, and response measures, but a multiplicity strategy is not reported. LOCF and pre-rescue substitution were used for the primary endpoint. The original sample-size rationale, interim-analysis plan, and sensitivity analyses are not reported in the registry record.