Comparison of Type 2 Diabetes Pharmacotherapy Regimens: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Comparison of Type 2 Diabetes Pharmacotherapy Regimens (NCT05073692), a registered phase not stated study. The primary outcome is Incidence of 3-point Major Adverse Cardiovascular Events (MACE).

2. Clinical Question

The study summary frames the question broadly: which glucose-lowering medications have favorable effects on heart health and other patient-important outcomes, when medications should be initiated, and whether effects differ across groups defined by characteristics such as age, sex, race or ethnicity, baseline heart health, or renal function. These aims describe the motivation, not results. The registry record does not report subgroup findings or an analysis of treatment timing. Nor does it identify the precise population used for each comparison. The listed enrollment should therefore not be treated as the denominator for any particular treatment group or outcome analysis.

3. Trial Design

The record identifies the study as “Comparison of Type 2 Diabetes Pharmacotherapy Regimens Using Targeted Learning” and lists Kaiser Permanente as the sponsor. Its status is completed. The listed start date is 2021-07-01, and the listed primary completion date is 2024-12-31. Enrollment is listed as 241981. The registry record does not describe whether this enrollment figure is an actual or estimated total.

The listed interventions are drug regimens or drug classes: SU, DPP4, SGLT2i, GLP-1RA, and SGLT2i or GLP-1RA. Specific agents named are Linagliptin (DPP4), Exenatide (GLP-1RA), and Liraglutide (GLP-1RA). These entries do not, by themselves, define mutually exclusive treatment groups, the time at which treatment is assigned, or the rules for switching and continuing medication. Those design details are not reported in the registry record.

The record explicitly states that this is not a randomized comparison, that allocation is not stated, and that the listed arms are not a like-for-like control comparison. It therefore does not support describing the groups as randomized arms or interpreting them as directly exchangeable. The official title refers to targeted learning, but a detailed statistical analysis plan, including the specific estimand and adjustment strategy, is not reported in the registry record.

One primary outcome measure is listed, with a timeframe of 2.5 years. The record lists no analyses and does not post primary outcome results. It also does not report the number of participants assigned to each regimen, eligibility criteria, follow-up completeness, or how treatment exposure and outcome timing were handled. These omissions limit what can be concluded about the design and the interpretation of a comparison, even though the registry status is completed.

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 Incidence of 3-point Major Adverse Cardiovascular Events (MACE), measured over 2.5 years.

The primary endpoint is incidence of 3-point major adverse cardiovascular events, or MACE, measured as a count of participants over 2.5 years. The registry defines this composite as nonfatal myocardial infarction, nonfatal stroke, or cardiovascular disease death. A participant experiencing one of the listed components meets the composite outcome definition; the record does not report whether component-specific results were also analyzed.

Composite endpoints are used in clinical research to capture a set of related outcomes under one outcome definition. In general, combining events can represent a broader clinical question than tracking only one event type. It can also make interpretation less direct: a composite result does not establish whether each component contributed similarly, or whether any apparent pattern was driven by one component. Those distinctions require component-specific data, which are not reported in the registry record.

The measure is specified as a count of participants with an event, not as a rate or a time-to-event estimate. A count alone does not express risk without a suitable denominator and information about follow-up. The registry record does not report the endpoint counts by group, denominators, comparative estimates, or confidence intervals. Thus the definition and timeframe are available, but the posted record does not show how often MACE occurred or how precisely any difference between regimens was estimated.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The official title names targeted learning, but the specific statistical procedures are not reported in the registry record. The explanations below are general descriptions of concepts relevant to interpreting a nonrandomized treatment comparison; they do not establish which procedures were used in this study.

What does a treatment comparison estimate?

General: a statistical comparison requires a clearly defined question about which treatment strategies are being contrasted, in which population, and over what follow-up period. It also requires a defined outcome and a measure of effect. For a binary outcome, possible summaries include the proportion of participants with an event in each group or a contrast between those proportions. A count of participants with an event is not itself a comparative effect unless it is interpreted with group denominators and a specified comparison. The registry identifies a count-based primary measure and a 2.5 years timeframe, but it does not report the comparison estimate or denominators.

What is targeted learning?

General: targeted learning is a family of statistical approaches designed to estimate a specified quantity from data while using models for relevant features of the data. Depending on the question and implementation, an analysis may use information about treatment assignment, outcome risk, and measured participant characteristics. The method's name alone does not reveal the target quantity, covariates, model choices, diagnostics, or assumptions. Those details are not reported in the registry record. It is therefore not possible to assess whether the analysis targeted a particular population average, how it handled treatment groups, or how uncertainty was estimated.

Why does the comparison structure matter?

General: random assignment can balance measured and unmeasured characteristics on average, whereas a nonrandomized comparison may retain differences between treatment groups. If factors that influence treatment choice also relate to the outcome, a crude comparison can reflect those differences as well as any treatment-related contrast. Adjustment for measured factors may reduce some confounding, but it depends on having relevant variables measured adequately and on correctly specifying the analysis. The registry explicitly says the comparison is not randomized, allocation is not stated, and the groups are not a like-for-like control comparison. It does not report the adjustment variables or methods, so the extent to which these issues were addressed cannot be judged.

How should uncertainty be read?

General: a confidence interval describes a range of estimates compatible with the data under the analysis assumptions and procedure. A narrow interval indicates greater statistical precision than a wide interval, but neither width nor interval location resolves systematic bias from design or measurement. A p-value, when reported, addresses compatibility with a specified null hypothesis under its assumptions; it does not measure the probability that a treatment works or establish the importance of an effect. No confidence intervals or p-values for the primary comparison are reported in the registry record.

What does the binary outcome imply?

General: a binary outcome records whether each participant meets the event definition during the relevant observation period. When the published measure is a count, interpretation depends on knowing who was at risk, how many participants were observed, and how missing follow-up was handled. If observation time varies, a simple count may not convey that variation. The registry identifies the outcome as a participant count and specifies 2.5 years as the timeframe, but it does not explain follow-up handling or provide analysis denominators.

What can planning illustrations show?

General: sample-size planning illustrations describe what sample sizes might be needed under stated assumptions about an effect and a statistical procedure. They are not observed power and cannot replace a reported estimate with its uncertainty. The registry's planning assumptions describe a two-sided significance level of 0.05, equal allocation, a normal approximation, and an effect specified as a fraction of the observed effect. The registry record does not report the observed effect or a primary analysis from which to assess it. Planning illustrations therefore cannot be read as evidence that a comparison was detected or that its result is precise.

Because the analysis details and primary results are absent, these general concepts clarify what information would matter but cannot be used to reconstruct the study's findings. No unreported method or estimate can be inferred from the title or the listed outcome definition.

10. Exact Estimates for Each Arm

No two-arm comparison is shown for this trial, but each arm's result can still be summarized with exact interval estimates computed from the posted counts. The Clopper-Pearson interval is exact and conservative; the Wilson interval is a common closer-to-nominal alternative; the Bayesian credible interval uses a uniform Beta(1, 1) prior. For small groups the three can differ noticeably.

R code and output

arm    <- c("SU Initiators", "DPP4 Initiators", "SGLT2i Initiators", "GLP-1RA Initiators")
n      <- c(212042, 5862, 14858, 9219)
events <- c(11348, 393, 424, 296)
for (i in seq_along(n)) {
  ci <- binom.test(events[i], n[i])$conf.int          # Clopper-Pearson exact interval
  cr <- qbeta(c(0.025, 0.975), 1 + events[i], 1 + n[i] - events[i])   # Bayesian, uniform Beta(1,1) prior
  wi <- (function(x, n) { p <- x / n; z <- qnorm(0.975); d <- 1 + z^2 / n
    c((p + z^2 / (2 * n) - z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d, (p + z^2 / (2 * n) + z * sqrt(p * (1 - p) / n + z^2 / (4 * n^2))) / d) })(events[i], n[i])
  cat(sprintf("ARM|%s|%d|%d|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f|%.1f\n", arm[i], n[i], events[i], 100 * events[i] / n[i],
              100 * ci[1], 100 * ci[2], 100 * wi[1], 100 * wi[2], 100 * cr[1], 100 * cr[2]))
}

ARM  SU Initiators  212042  11348  5.4  5.3  5.4  5.3  5.4  5.3  5.4
ARM  DPP4 Initiators  5862  393  6.7  6.1  7.4  6.1  7.4  6.1  7.4
ARM  SGLT2i Initiators  14858  424  2.9  2.6  3.1  2.6  3.1  2.6  3.1
ARM  GLP-1RA Initiators  9219  296  3.2  2.9  3.6  2.9  3.6  2.9  3.6

11. Multiplicity, Missing Data and Analysis Populations

The registry lists one primary outcome measure, incidence of 3-point MACE. It does not report a detailed plan for handling multiplicity across treatment comparisons, other outcomes, or subgroup questions. The study summary names several potential sources of variation in treatment effects, including age, sex, race or ethnicity, baseline heart health, and renal function, but the record does not say whether these were prespecified analyses, how many comparisons were planned, or how multiplicity was addressed. Without that information, a reader cannot assess the risk of chance findings arising from multiple testing.

Missing-data methods are not reported in the registry record. In general, missing outcome information can affect estimates if the availability of follow-up differs across participants or is related to their treatment or outcome. Appropriate interpretation depends on knowing how missing observations were identified and handled, and whether any assumptions or sensitivity analyses were used. The record does not report follow-up completeness, missingness by group, or methods for addressing missing data.

The analysis populations are also not reported in the registry record. It does not state which participants were eligible for a particular comparison, whether inclusion depended on treatment initiation or continued exposure, or how participants with incomplete information were handled. Nor does it provide group-specific analysis denominators. The listed enrollment of 241981 cannot establish how many participants were included in each analysis. With no posted analyses, results, or analysis-population definitions, the robustness of any comparison to exclusions, missingness, or multiple testing cannot be evaluated.

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

The registry record means that a completed study was registered with a stated goal of comparing type 2 diabetes pharmacotherapy regimens using targeted learning. It names a primary endpoint: participants with 3-point MACE, defined as nonfatal myocardial infarction, nonfatal stroke, or cardiovascular disease death, over 2.5 years. It also describes broad interests in treatment timing and whether effects might differ across participant characteristics.

It does not mean that the record demonstrates a difference between the listed regimens. No primary outcome counts or comparative estimates are reported, and no analyses are listed. There is no posted measure of the size of an effect or its statistical precision. The record cannot establish whether a regimen was associated with more or fewer events, whether estimates were consistent across groups, or whether a result would be clinically important.

The registry's nonrandomized-comparison note also matters for interpretation. General statistical reasoning indicates that differences between groups can reflect characteristics associated with treatment choice as well as treatment itself. The record does not describe the analysis well enough to determine how such differences were handled. A method named in the official title does not by itself establish that confounding was removed or that the comparison supports a causal conclusion.

The listed deaths in adverse-event information are not a substitute for the primary outcome analysis. The registry does not present them as MACE outcomes or provide the denominators and comparative context needed to interpret them as rates. Similarly, the enrollment figure does not indicate the number analyzed in each group or the precision of an estimate. The record is therefore informative about the study's stated purpose, listed interventions, and endpoint definition, but not about comparative clinical effects.

13. Why This Trial Matters Statistically

The study raises a statistically important question: how to compare medication regimens when treatment allocation is not randomized and the groups are not described as like-for-like controls. Such comparisons may address questions that are difficult to answer through a randomized design, but their interpretation depends on clear treatment definitions, a well-specified target population, measured confounders, and transparent analysis choices. The registry identifies targeted learning in the official title, yet does not report those analytical details.

The primary outcome also illustrates a key issue in clinical statistics. MACE is a composite of nonfatal myocardial infarction, nonfatal stroke, and cardiovascular disease death. A composite can represent several related outcomes in a single measure, but readers need the definition and, where available, component-level findings to understand what contributes to a result. The registry defines the components but posts no outcome estimates or component-specific results.

The record's broad aims include treatment timing and possible differences across participant characteristics. Those are distinct statistical questions that require clear definitions and analyses; they cannot be inferred from an overall endpoint description. Without posted analyses, readers cannot assess how the study addressed these questions or how precise any estimates were.

This record is a useful reminder that a completed registration and a substantial listed enrollment do not themselves communicate study findings. Interpretation requires results, denominators, uncertainty estimates, and enough method detail to understand the comparison. Those elements are not reported for the primary outcome in the registry record, so the statistical importance lies chiefly in recognizing what the study set out to investigate and what remains unavailable for evaluation.

14. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The registry record leaves important design and reporting details unavailable. It states that the comparison is not randomized, that allocation is not stated, and that the listed arms are not a like-for-like control comparison. It does not explain how participants entered treatment groups, define the treatment strategies in operational terms, or report whether treatment could change during follow-up. Those details are important for understanding the meaning of a regimen comparison.

The record does not report eligibility criteria, participant characteristics, group-specific enrollment, outcome counts, denominators, or follow-up completeness. Although enrollment is listed as 241981, that figure does not show how many participants were included in each analysis. The primary outcome is specified as a count of participants over 2.5 years, but the record does not explain the handling of unequal observation time, censoring, or missing outcome data.

No primary outcome estimates, confidence intervals, p-values, or analyses are listed. The official title names targeted learning, but the estimator, target quantity, covariates, and uncertainty method are not reported. The registry summary also raises questions about timing and subgroup variation without reporting corresponding analysis results. Multiplicity handling and analysis populations are not described.

The adverse-event information lists deaths for several named groups, but does not provide the denominators or context required to interpret those counts comparatively. The record does not identify them as primary MACE results. Taken together, these gaps prevent evaluation of the magnitude, precision, and robustness of a treatment comparison from this registry record alone.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Comparison of Type 2 Diabetes Pharmacotherapy Regimens — NCT05073692.
• Registry record last updated 2025-11-21; status last verified 2025-11.
• Results first posted 2025-11-21.
• 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.

18. Record Summary

The registry record for NCT05073692 is titled “Comparison of Type 2 Diabetes Pharmacotherapy Regimens” and lists the official title “Comparison of Type 2 Diabetes Pharmacotherapy Regimens Using Targeted Learning.” Kaiser Permanente is the listed sponsor. The study status is completed, with a listed start date of 2021-07-01, primary completion date of 2024-12-31, and enrollment of 241981.

Listed interventions include SU, DPP4, SGLT2i, GLP-1RA, SGLT2i or GLP-1RA, Linagliptin (DPP4), Exenatide (GLP-1RA), and Liraglutide (GLP-1RA). The primary measure is incidence of 3-point MACE over 2.5 years, defined as nonfatal myocardial infarction, nonfatal stroke, or cardiovascular disease death.

The record states that this is not a randomized comparison, allocation is not stated, and the groups are not a like-for-like control comparison. It lists no analyses and posts no primary outcome results. Comparative effects, their precision, and the methods used to address confounding, missing data, or multiplicity are not reported in the registry record.