Cardiovascular Outcomes in Bariatric Surgery Patients With Type 2 Diabetes: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Cardiovascular Outcomes in Bariatric Surgery Patients With Type 2 Diabetes (NCT03955952), a registered phase not stated study. The primary outcome is All-Cause Mortality and/or Cardiovascular Outcomes.

2. Clinical Question

This is a question about an association between surgery and a composite outcome, not a randomized treatment effect. The registry record does not report a primary-outcome comparison or an estimate of the association. It therefore does not establish whether the recorded outcome differed between groups, or how precisely any difference was estimated.

3. Trial Design

NCT03955952 is listed as COMPLETED and sponsored by The Cleveland Clinic. The registry record gives enrollment as 91700. The listed start date is 2019-06-05, and the primary completion date is 2019-07-12. These registry entries describe the record but do not give a full account of how participants were identified, followed, or analyzed.

The comparison is between patients with type 2 diabetes and BMI >=30 who underwent bariatric surgery and matched non-surgical controls with type 2 diabetes and BMI >=30. The record explicitly notes that allocation was not stated, that the comparison was not randomized, and that the arms are not a like-for-like control comparison. It does not describe the matching procedure, which characteristics were matched, or whether matching was incorporated into the outcome analysis. Those details are not reported in the registry record.

The primary endpoint is a composite, recorded as a percentage of participants, and its timeframe runs from index date through a maximum of 8 years. The registry describes the endpoint as the first occurrence of all-cause mortality, coronary artery disease event, stroke, heart failure, nephropathy, or atrial fibrillation. The record also lists secondary outcomes including Major Adverse Cardiovascular Event (MACE-3), all-cause mortality, coronary disease events, cerebrovascular events, heart failure, development of nephropathy, and atrial fibrillation.

The registry lists 8 outcome measures and 0 analyses. Outcome estimates, methods of adjustment, and details of follow-up and censoring are not reported in the registry record. Consequently, the listed design and endpoint descriptions do not supply the information needed to reproduce an outcome comparison or judge its precision.

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

Analysis population for the primary outcome, as defined in the registry: All enrolled participants

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 All-Cause Mortality and/or Cardiovascular Outcomes, measured over From index date through a maximum of 8 years.

The primary endpoint combines the first occurrence of all-cause mortality or a cardiovascular or related event: coronary artery disease event, stroke, heart failure, nephropathy, or atrial fibrillation. Its stated unit is the percentage of participants with an event, and the timeframe is from index date through a maximum of 8 years. This means the endpoint is intended to count a participant when any qualifying component occurs first, rather than to count every later event for that participant.

Composite endpoints are used in clinical research to bring several clinically important outcomes into a shared measure. In general, that can capture a broader range of outcomes than any single component and can make a study question more inclusive. Interpretation depends on the components: a composite result does not by itself show which component contributes most, whether the components have similar importance, or whether each component follows the same pattern.

The registry record defines the component events but does not report their individual results or the composite result. It also does not report an endpoint estimate, confidence interval, or a comparison between the groups. The label should therefore be read as a planned or specified outcome definition, not as evidence that the endpoint occurred at a particular rate or differed between arms.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry identifies the primary outcome as a percentage of participants with an event, with a 95% Confidence Interval as the dispersion measure. It does not report an actual percentage or confidence interval, and it lists 0 analyses. The method descriptions below explain the statistical ideas relevant to the stated endpoint; they do not establish which analysis was carried out.

What does a percentage of participants represent?

In general, a percentage for a binary outcome represents the proportion of participants classified as having an event under a defined outcome rule. Here, the rule is the first occurrence of any listed endpoint component during follow-up from index date through a maximum of 8 years. The denominator and event count are both needed to interpret such a percentage. The registry record does not report them for the primary endpoint, by group or overall.

For a time-bounded outcome, interpretation also depends on how follow-up is handled. Participants may have different amounts of observation, and some may not have a recorded event before their observation ends. Time-to-event methods can account for timing and censoring under stated assumptions; a simple percentage generally needs a clearly defined follow-up window and denominator. The record does not report the primary analysis method, censoring rules, or participant follow-up distributions. It is therefore not possible to determine from the registry record how the endpoint timeframe was translated into an analysis.

What does a confidence interval add?

In general, a confidence interval conveys statistical uncertainty around an estimate. A narrower interval indicates greater precision under the method used, while a wider interval indicates less precision. It does not remove possible bias in the underlying comparison, and it does not show which endpoint components drove a composite result. Although the registry names a 95% confidence interval as the dispersion measure, no interval is reported for the primary endpoint. The precision of the primary result is therefore unavailable.

Why does the comparison design matter?

In a randomized comparison, allocation is used to create groups whose measured and unmeasured characteristics are expected to be balanced on average. This record does not describe a randomized comparison; it states that allocation was not stated and identifies matched non-surgical controls. Matching can improve comparability on the characteristics used to form matched groups, but it does not automatically ensure comparability on other characteristics. In general, outcome analyses of matched data may also need to account for the matching structure. The registry record does not report the matching variables, how matching was performed, or whether the analysis accounted for it.

In nonrandomized comparisons, confounding is a general concern: characteristics related to both treatment selection and outcome can contribute to an observed association. Adjustment methods can address measured characteristics when they are adequately recorded and modeled, but cannot guarantee removal of unmeasured confounding. The registry record does not report covariate adjustment, model specification, or sensitivity analyses. No particular adjustment strategy can be attributed to this study from the posted record.

How should planning illustrations be understood?

The registry describes sample-size illustrations as planning calculations based on the observed effect, not as the trial's own design. The stated assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation; illustrations use a stated fraction of the observed effect to detect. Such calculations are hypothetical planning aids. They are not observed power, do not indicate the probability that the reported result is true, and do not replace the outcome estimate or its uncertainty. The record does not report the trial's own sample-size justification.

The record distinguishes time-to-event illustrations, which give a total number of events under the Schoenfeld approach, from other illustrations, which give participants per arm. This distinction is general to the listed planning approach; it does not supply a result for the primary endpoint. The posted record does not include enough primary-outcome results to assess whether any such illustration corresponds to the achieved data.

What is not documented?

The analysis method, model assumptions, treatment of repeated or competing events, covariate selection, and handling of missing data are not reported in the registry record. These omissions matter because choices about endpoint construction, follow-up, and adjustment can affect both the estimate and its precision. Without those details and the numerical outcome results, a reader cannot reproduce the primary analysis or evaluate whether its statistical assumptions were appropriate.

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("Bariatric Surgery", "Non-Surgical Control")
n      <- c(2287, 11435)
events <- c(704, 5454)
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  Bariatric Surgery  2287  704  30.8  28.9  32.7  28.9  32.7  28.9  32.7
ARM  Non-Surgical Control  11435  5454  47.7  46.8  48.6  46.8  48.6  46.8  48.6

11. Multiplicity, Missing Data and Analysis Populations

The registry lists 8 outcome measures, including the primary composite and secondary outcomes such as MACE-3, all-cause mortality, coronary disease events, cerebrovascular events, heart failure, development of nephropathy, and atrial fibrillation. When several outcomes or comparisons are examined, the chance of at least one apparently notable result can increase. In general, prespecifying a primary endpoint and setting out how multiple tests will be interpreted can help readers distinguish confirmatory from exploratory findings.

The registry record does not report a multiplicity strategy, such as an adjustment procedure, hierarchy among secondary outcomes, or a distinction between confirmatory and exploratory analyses. It also lists 0 analyses and provides no outcome estimates or p-values. The statistical handling of multiple outcomes is therefore not reported in the registry record, and no claim about adjustment or statistical significance can be made.

The primary endpoint population is described as all enrolled participants. That label does not establish which participants were included in a completed analysis or how missing outcome information was handled. The registry record does not report missing-data procedures, reasons for missingness, exclusions from analysis, or an analysis-population definition beyond the endpoint description. It also does not say whether an intention-to-treat, per-protocol, or other analysis population was used. These details are not reported in the registry record.

In general, missing outcome data can affect estimates when the likelihood of being observed is related to participant characteristics or outcomes. The impact depends on the amount, pattern, and handling of missingness. Without those details, the direction or magnitude of any potential effect cannot be assessed for this comparison.

12. 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 0% reporting threshold.

Most frequent serious adverse event terms

Serious events with the highest arm-level rate: bleeding requiring transfusion (3.0%), pulmonary adverse events (2.5%), venous thrombosis (1.0%), cardiac events (0.7%), renal failure requiring dialysis (0.2%).

The registry's adverse-event fields list, for Bariatric Surgery, 2287 participants at risk for serious events and 169 affected; the other-event fields are 0 at risk and 0 affected, and the death count is 112. For Non-Surgical Control, the serious-event and other-event fields are all 0, and the death count is 1111. These entries are not accompanied by a clear explanation of the denominators or how they relate to the endpoint population. They should be read as registry fields, not as a complete group-level safety comparison.

The record also lists serious adverse-event terms with percentages: Bleeding requiring transfusion, 3.0%; Pulmonary Adverse Events, 2.5%; Venous Thrombosis, 1.0%; Cardiac events, 0.7%; and Renal Failure requiring dialysis, 0.2%. The record does not identify the relevant arm, denominator, observation period, or whether these terms overlap. Those percentages cannot be interpreted as comparative risks without that context.

In general, safety counts and percentages need a defined population, exposure period, event definition, and denominator. The posted adverse-event details do not provide enough information to reconcile the listed fields or compare groups reliably. They are also not a substitute for the primary composite outcome analysis.

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

The record specifies a question about the relationship between bariatric surgery and a composite of all-cause mortality and cardiovascular or related outcomes among patients with type 2 diabetes and obesity. It describes the surgery group, matched non-surgical controls, the endpoint components, and a maximum follow-up timeframe of 8 years. Those details help clarify what the study set out to examine.

The record does not provide the primary endpoint percentage in either group, a between-group estimate, or a 95% confidence interval. It therefore does not show the direction or size of any association, nor how precisely one was estimated. The secondary outcomes are listed, but their results are not reported. No conclusion about an observed difference in the primary or secondary outcomes can be drawn from the registry information.

The nonrandomized nature of the comparison is important to interpretation. The registry says allocation was not stated and that the arms are not a like-for-like control comparison. Even if group outcome estimates were available, observed differences would need to be interpreted in light of the groups' comparability, the matching process, and any adjustment used. The record does not report enough detail to evaluate these aspects.

In general, a statistical association is not automatically a causal effect. A result from a nonrandomized comparison may be influenced by measured or unmeasured differences between groups. Conversely, the absence of a reported estimate is not evidence that there was no association. The registry record leaves the direction, magnitude, precision, and robustness of the primary comparison unresolved.

The listed adverse-event terms and counts also require context about denominators, arms, follow-up, and definitions. They should not be taken as evidence of a comparative safety effect or used to fill the gap in the primary endpoint results. The appropriate reading is limited: the record defines an analysis question and records some safety fields, but does not present enough results to determine what the study found statistically.

14. Why This Trial Matters Statistically

This record is statistically notable because it describes a large observational comparison while leaving the central outcome analysis unreported. Enrollment is listed as 91700, and the study question concerns a composite that includes mortality and several cardiovascular or related events. Yet the registry lists 0 analyses and gives no primary outcome estimate or confidence interval. This illustrates why enrollment size alone cannot tell a reader whether an estimate is available, precise, or interpretable.

The comparison also makes design information essential. The controls are described as matched, but allocation is not stated; the record explicitly cautions that the comparison is not randomized and the arms are not a like-for-like control comparison. Matching can be useful, but interpretation depends on how it was done and how the analysis uses it. Those details are not reported. For biostatisticians and clinical researchers, the case underscores the need to distinguish a description of a study question from a transparent account of its estimand, analysis population, model, and uncertainty.

The broad composite endpoint raises another general statistical consideration: an overall composite can conceal different patterns across its components. Component-specific results and clear outcome definitions help show what an aggregate estimate represents. Here, the component definitions are named, but their results are not reported.

Finally, registry fields should be read in context. The adverse-event counts and percentages lack enough denominator and group detail to support a reliable comparative interpretation. This record therefore demonstrates the importance of complete reporting: design, outcome estimates, uncertainty, analysis methods, and safety denominators are all needed to move from a stated question to an interpretable statistical result.

15. 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 does not report primary endpoint results, including group-specific percentages, a comparative estimate, or a 95% confidence interval. It lists 0 analyses, so the posted record does not permit an assessment of the endpoint's direction, magnitude, or precision.

The study is identified as a nonrandomized comparison, allocation is not stated, and the record cautions that the arms are not a like-for-like control comparison. Although the control group is described as matched, the matching variables, matching procedure, and handling of matched data in analysis are not reported. Covariate adjustment and sensitivity analyses are also not reported in the registry record.

The endpoint timeframe is stated as from index date through a maximum of 8 years, but the record does not describe follow-up completeness, censoring, or how different observation times were handled. It does not report primary outcome denominators, missing-data procedures, or the analysis population beyond the description of all enrolled participants. The listed enrollment cannot be assumed to equal the endpoint analysis population.

The record lists multiple outcome measures but does not report a plan for multiplicity or results for the secondary outcomes. It does not provide enough detail about the definition of MACE-3 or component-level results to interpret the composite's contribution structure. The adverse-event information also lacks clear denominators, arm attribution for the listed percentages, and relevant follow-up context. These limitations are specific reporting gaps in the registry record; their effects on any underlying analysis cannot be determined from the posted information.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Cardiovascular Outcomes in Bariatric Surgery Patients With Type 2 Diabetes — NCT03955952.
• Registry record last updated 2026-08-03; status last verified 2026-07.
• Results first posted 2026-08-03.
• 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.

19. Record Summary

NCT03955952, sponsored by The Cleveland Clinic, is listed as COMPLETED, with enrollment of 91700. The registry describes a comparison of bariatric surgery with matched non-surgical controls among patients with type 2 diabetes and BMI >=30. Allocation is not stated; the record notes that the comparison is not randomized and that the arms are not a like-for-like control comparison.

The primary endpoint is the percentage of participants with the first occurrence of all-cause mortality, coronary artery disease event, stroke, heart failure, nephropathy, or atrial fibrillation, from index date through a maximum of 8 years. The record lists 8 outcome measures and 0 analyses. It does not report primary endpoint estimates, confidence intervals, analysis methods, or complete context for interpreting the adverse-event fields. The registry therefore documents the study question and endpoint definition, but not a statistical result for the primary comparison.