This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial results on this page are restricted to the results posted in the ClinicalTrials.gov record; calculations added by Clinical Biostats are labelled as such.
1. Trial at a Glance
This page covers the registry results for Metformin in Heart Failure Without Diabetes (NCT03331861), a randomized phase 2 study. The primary outcome is Change in Minute Ventilation to Carbon Dioxide Production (VE/VCO2) Slope.
| Feature | Registry value |
|---|---|
| ClinicalTrials.gov identifier | NCT03331861 |
| Phase | Phase 2 |
| Condition | Heart Failure |
| Allocation | Randomized |
| Design model | Parallel |
| Masking | Quadruple |
| Enrollment | 6 |
| Arms in the primary outcome | n/a |
| Lead sponsor | Columbia University |
| Results posted | Yes |
2. Clinical Question
Population
People with heart failure who were not diabetic and had no evidence of insulin resistance, as described in the study summary. Other eligibility criteria are not reported in the registry record.
Intervention
Metformin hydrochloride, described as metformin for 6 months.
Comparator
Placebo, described as a matched placebo for 6 months.
Primary question
The planned primary outcome was change in the minute ventilation to carbon dioxide production slope from baseline to 6 months. Because the study was terminated early and data were not collected, the record does not provide a comparison of that outcome between the groups.
The study was intended as a pilot clinical research study of metformin in heart failure without diabetes or evidence of insulin resistance. Random assignment and masking were part of the registered design, but the record does not describe the randomization procedure or which people were masked. It also does not report an allocation ratio.
The clinical question was whether the planned change in the VE/VCO2 slope could be assessed in this population under a metformin-versus-placebo comparison. The registry also lists changes in VAS score, MLHFQ score, peak myocardial oxygen consumption, and heart failure hospitalizations as secondary outcomes. These are outcome plans, not findings: the record says outcome data were not collected due to low enrollment. Consequently, it cannot establish whether groups differed on the primary or secondary outcomes, or how precisely any difference might have been estimated.
3. Trial Design
The registry describes a randomized, quadruple-masked, parallel-group phase 2 study sponsored by Columbia University. The study compared metformin hydrochloride with matched placebo, with both interventions described as lasting 6 months. The registry lists heart failure as the condition and describes the intended population as people without diabetes and without evidence of insulin resistance.
The recorded enrollment is 6. The study began on 2019-02-15, had a primary completion date of 2021-08-20, and is listed as terminated. The registry record states that the study was terminated early and that data were not collected because enrollment was low. It therefore does not contain an analyzed treatment comparison for the primary outcome.
Randomization is intended to reduce systematic differences between assigned groups, while masking can reduce the influence of treatment knowledge on participant experience, study conduct, or outcome assessment. Those are general design aims; they do not substitute for evidence that the study achieved comparable groups or produced interpretable outcome data. Details about the randomization method, concealment, masking roles, and any departures from the registered design are not reported in the registry record.
The registry lists the primary outcome as change in the VE/VCO2 slope between baseline and 6 months. It also identifies secondary outcomes involving VAS score, MLHFQ score, peak myocardial oxygen consumption, and heart failure hospitalizations. Although the registry records 5 outcome measures, it records 0 analyses. The information available therefore describes the planned design and outcomes, but not an efficacy analysis or an estimate of the difference between metformin and placebo.
The record does not report a sample-size justification or a statistical analysis plan. The termination and lack of collected outcome data are central to interpreting this trial: randomization and masking describe intended methods, but no primary-outcome comparison is available from which to assess the intervention groups.
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: 2 out of 6 participants screen failed and no further information was collected or analyzed.
Overall Study
| Stage | Metformin | Placebo |
|---|---|---|
| STARTED | 2 | 2 |
| COMPLETED | 2 | 2 |
| NOT COMPLETED | 0 | 0 |
Analysis population for the primary outcome, as defined in the registry: Study was terminated early; data was not collected due to low enrollment, thus cannot be analyzed.
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.
| Characteristic | Unit | Metformin | Placebo | Total |
|---|---|---|---|---|
| Participants analysed | 2 | 2 | 4 | |
| Age, Categorical: <=18 years | Participants | 0 | 0 | 0 |
| Age, Categorical: Between 18 and 65 years | Participants | 2 | 1 | 3 |
| Age, Categorical: >=65 years | Participants | 0 | 1 | 1 |
| Sex: Female, Male: Female | Participants | 1 | 1 | 2 |
| Sex: Female, Male: Male | Participants | 1 | 1 | 2 |
| Ethnicity (NIH/OMB): Hispanic or Latino | Participants | 0 | 1 | 1 |
| Ethnicity (NIH/OMB): Not Hispanic or Latino | Participants | 2 | 1 | 3 |
| Ethnicity (NIH/OMB): Unknown or Not Reported | Participants | 0 | 0 | 0 |
| Race (NIH/OMB): American Indian or Alaska Native | Participants | 0 | 0 | 0 |
| Race (NIH/OMB): Asian | Participants | 0 | 0 | 0 |
| Race (NIH/OMB): Native Hawaiian or Other Pacific Islander | Participants | 0 | 0 | 0 |
| Race (NIH/OMB): Black or African American | Participants | 0 | 1 | 1 |
| Race (NIH/OMB): White | Participants | 2 | 1 | 3 |
| Race (NIH/OMB): More than one race | Participants | 0 | 0 | 0 |
| Race (NIH/OMB): Unknown or Not Reported | Participants | 0 | 0 | 0 |
| Region of Enrollment: United States | participants | 2 | 2 | 4 |
6. Primary Endpoint
The registry lists the primary outcome as Change in Minute Ventilation to Carbon Dioxide Production (VE/VCO2) Slope, measured over Baseline and 6 months.
| Element | Registry entry |
|---|---|
| Result type | N/A |
| Measure of spread | not applicable |
| Unit | not stated |
| Endpoint classification | not reduced to a single comparison |
The primary endpoint is change in the minute ventilation to carbon dioxide production slope, abbreviated VE/VCO2 slope, from baseline to 6 months. Minute ventilation describes the volume of air moved in and out of the lungs over time; carbon dioxide production describes the rate at which carbon dioxide is produced. The slope summarizes how ventilation changes in relation to carbon dioxide production during the assessment. A change in that slope was the registered primary measure for comparing the study groups over the planned follow-up.
As a general measurement principle, a slope can condense a relationship across multiple paired observations into a single summary. That makes it possible to define a change score for each participant and compare those changes between groups, provided the measurements and analysis are available. The record does not specify the endpoint unit, the calculation details beyond stating that the slope change would be calculated, or a clinically important threshold for interpreting its size.
The registry description mentions normality assessment for VO2 change outcomes using Q-Q plots, histograms, and Kolmogorov-Smirnov tests. It also says the VE/VCO2 slope change would be calculated. The wording does not establish that an endpoint value was obtained or analyzed. The record states that data were not collected, so it gives no observed change, group estimate, confidence interval, or test result for the primary endpoint.
The endpoint was selected to assess a physiological measure in the planned comparison; that choice is not itself evidence of a treatment effect. Without collected observations, the endpoint cannot distinguish between a difference, no difference, or uncertainty too great to draw a conclusion.
7. Why No Re-analysis Is Shown
8. Statistical Interpretation
The main statistical finding is the absence of an analyzable primary-outcome result. The registry states that the study was terminated early and data were not collected due to low enrollment. It records 0 analyses. Thus, there is no posted estimate of the change in VE/VCO2 slope for either group, no between-group effect estimate, and no confidence interval or p-value for the primary comparison.
This absence must not be read as a finding that the groups were equal or that the intervention had no effect. An estimate of zero would be a result calculated from observed data; no such estimate is reported. Likewise, without outcome observations, the record cannot show whether the data were compatible with benefit, harm, or no meaningful difference. The study did not produce the information needed to make those statistical distinctions.
The registry lists secondary outcomes for VAS score, MLHFQ score, peak myocardial oxygen consumption, and heart failure hospitalizations. These are registered outcomes, not reported results. The absence of analysis means there is no basis in this record to describe their group differences or uncertainty. No outcome-specific analysis populations or missing-data handling procedures are reported.
The safety record lists 2 participants at risk in each group. It records 0 serious adverse events, 0 other adverse events, and 0 deaths in the Metformin group, and the same reported counts in the Placebo group. These figures are descriptive counts in the registry, not an estimate of comparative safety. The record does not provide an analysis of risk or precision, and the small reported at-risk groups and absence of efficacy data limit what can be inferred.
In general, a confidence interval indicates the range of effect estimates compatible with the data and the statistical method used; its width conveys precision. A p-value, when reported, describes how unusual data at least as incompatible with a specified null hypothesis would be under that hypothesis and the model assumptions. Neither measure can be recovered from the registry's statement that outcome data were not collected. They should not be inferred from enrollment, safety counts, or the planned design.
The enrollment of 6 is the recorded trial enrollment, not a substitute for an outcome analysis. The registry does not report how many participants, if any, contributed measurements at each planned assessment, and it states that data were not collected. No treatment-effect conclusion or precision assessment is supported by the posted record.
9. Statistical Methods Explained
The registered design combines random assignment, masking, and parallel groups. In general, random assignment aims to make allocation independent of participant characteristics, so that systematic differences are less likely to explain a treatment contrast. Masking is intended to limit the influence of knowledge about assignment. A parallel-group design compares separate groups over the study period rather than switching the same participants between interventions. These are design principles; the registry does not report operational details or outcome analyses demonstrating how they functioned in this study.
What does the primary outcome summarize?
The primary measure is change in the VE/VCO2 slope between baseline and 6 months. A slope describes the relationship between two measured quantities; here, the named quantities are minute ventilation and carbon dioxide production. The registry says the slope change would be calculated, but does not provide the calculation procedure, units, or observed values. As a general statistical approach, a participant-level change score can be compared across randomized groups, but such a comparison requires collected measurements. The record explicitly says data were not collected, so no change-score comparison is available.
Why assess distributions?
The endpoint description refers to assessing normality for VO2 change outcomes with visual assessments using Q-Q plots and histograms, as well as Kolmogorov-Smirnov tests. These methods address whether a distribution appears compatible with a normal model. A Q-Q plot compares ordered observations with the pattern expected under a reference distribution; a histogram displays the observed distribution's shape. A Kolmogorov-Smirnov test provides a formal comparison with a specified distribution. As a general principle, distributional checks can help identify strong departures from assumptions behind some methods, but a test result does not prove that a model is appropriate or inappropriate in every respect. The registry does not report test results, and the stated lack of collected data means these checks cannot be interpreted as performed analyses.
The description refers to VO2 change outcomes when discussing normality, while the listed primary outcome is VE/VCO2 slope change. It does not explain the relationship between those statements or provide a full analysis plan. The record says the slope change would be calculated, but does not report how a group contrast would be estimated, which model would be used, or how departures from distributional assumptions would be handled.
How should estimates and uncertainty be read?
Generally, a treatment-effect estimate describes the observed contrast between groups on a defined outcome. A confidence interval communicates uncertainty around that estimate under the chosen method and assumptions. A narrow interval indicates greater precision than a wide interval, while an interval spanning values in different directions may leave the direction or magnitude uncertain. These concepts apply only when data and an analysis are available. No primary effect estimate or confidence interval is reported in the registry.
Similarly, a p-value is not a measure of clinical importance and does not state the probability that a treatment works. It is interpreted relative to a specified statistical model and null hypothesis. The record gives no primary-outcome p-value and records 0 analyses. The significance level 0.05 appears in the assumptions for a planning illustration; it is not a reported primary-outcome test result and should not be presented as the trial's observed statistical finding.
What do the planning assumptions mean?
The registry's sample-size assumptions describe a planning illustration based on a stated fraction of an observed effect, a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. These are assumptions for the illustration, not evidence that the trial used them to set its enrollment target. The registry does not report a sample-size justification or protocol rationale for enrollment. Because the trial record says outcome data were not collected, the planning assumptions cannot be interpreted as evidence of an observed treatment effect or observed power. Sample-size planning concerns the ability of a future design to detect an assumed effect under specified assumptions; it does not establish what this trial found.
In general, over-dispersion occurs when count data vary more than a chosen model expects. It can make standard errors from an unsuitable model too small and uncertainty appear less than it is. This is a general issue for some count outcomes, including event counts, not a reported analysis method in this registry. The record does not report a model for hospitalizations, an event-rate comparison, or any adjustment for over-dispersion.
What can be concluded from the recorded analyses?
The registry records 5 outcome measures and 0 analyses. That distinction matters: listing an outcome does not mean it was measured, and a planned method does not mean it was carried out. The record states that data were not collected due to low enrollment after early termination. Consequently, the endpoint descriptions and distributional procedures convey planned measurement and assessment, while statistical estimates, tests, model diagnostics, and sensitivity analyses are not reported in the registry record.
No recalculated treatment effect is appropriate without posted group summaries or outcome observations. Any sample-size illustration should be treated as planning information rather than a result. The sponsor's analysis cannot be compared with an unadjusted recalculation because no analysis or outcome summary is reported.
10. Multiplicity, Missing Data and Analysis Populations
The registry lists a primary outcome and secondary outcomes covering VAS score, MLHFQ score, peak myocardial oxygen consumption, and heart failure hospitalizations. Assessing multiple outcomes raises a general statistical issue: repeated testing can increase the chance that at least one result appears notable by chance if each test is interpreted separately. A prespecified hierarchy, adjustment procedure, or clearly defined exploratory interpretation can address that concern, but the registry record does not report a multiplicity strategy.
There are no outcome analyses reported, so there are no primary or secondary p-values to evaluate, adjusted or otherwise. The registry records 0 analyses and states that outcome data were not collected because the trial was terminated early for low enrollment. The listed secondary outcomes should therefore be understood as planned measures, not evidence of measured treatment differences.
The record does not report an analysis population, such as a population defined by randomized assignment or by available measurements. It also does not describe missing-data procedures, imputation, or sensitivity analyses. Since the registry says data were not collected, it is not possible to characterize outcome-level missingness, determine how many measurements would have been available at each time, or assess whether an analysis included participants according to their assigned group.
In general, missing-data methods rely on assumptions about why observations are absent and how their absence relates to measured or unmeasured values. Those assumptions cannot be assessed from the record. Here, the central limitation is not a reported statistical adjustment for missing endpoint observations; it is that outcome data were not collected. No imputed results or analysis population should be inferred.
11. Safety Results
The registry reports serious adverse events as affected participants divided by participants at risk.
| Arm | Participants with serious adverse events | Participants at risk | Percentage |
|---|---|---|---|
| Metformin | 0 | 2 | 0.0% |
| Placebo | 0 | 2 | 0.0% |
Deaths
| Arm | Deaths (participants affected of at risk) |
|---|---|
| Metformin | 0 of 2 |
| Placebo | 0 of 2 |
Other (non-serious) adverse events
Participants with at least one non-serious event at or above the registry's 5% reporting threshold.
| Arm | Participants affected of at risk |
|---|---|
| Metformin | 0 of 2 |
| Placebo | 0 of 2 |
The registry reports serious and other adverse-event figures separately for Metformin and Placebo. For each group, 2 participants were at risk. The Metformin group has 0 serious adverse events, 0 other adverse events, and 0 deaths recorded; the Placebo group has the same reported counts. The registry lists no top serious adverse-event terms.
These are reported counts, not comparative risk estimates. A count of 0 in a group does not establish that the intervention has no safety risk, and the record does not provide confidence intervals or a statistical comparison of adverse events. The figures also do not explain how adverse events were defined, collected, or monitored. Those details are not reported in the registry record.
Safety counts should be read alongside the stated early termination and lack of collected outcome data. The record does not explain how the reported safety at-risk groups relate to enrollment or provide further detail on safety follow-up. The counts describe what the registry reports; they do not support a broader conclusion about comparative safety or events outside the recorded categories.
12. What the Result Does, and Does Not, Mean
This registry record documents a randomized, quadruple-masked phase 2 pilot comparing metformin with matched placebo for heart failure in people described as not diabetic and without evidence of insulin resistance. It identifies change in VE/VCO2 slope from baseline to 6 months as the primary outcome and lists several secondary outcomes. Those details explain the intended question and design.
The record also states that the study was terminated early and data were not collected due to low enrollment. As a result, it does not report an effect estimate for the primary outcome, a comparison of secondary outcomes, or statistical uncertainty around any treatment difference. The absence of an analysis is not evidence of no effect, equal effects, or a particular direction of effect. It means the registry does not contain outcome results capable of answering the planned clinical question.
The safety figures record 0 serious adverse events, 0 other adverse events, and 0 deaths in each group, with 2 participants at risk in each. These are descriptive registry counts, not proof that an intervention is risk-free or a precise comparison between groups. Their interpretation is limited by the amount and detail of safety information reported.
The enrollment and design cannot compensate for the missing outcome measurements. Nor can the planning assumptions establish how the trial would have performed or what it would have found. The posted record is informative about the trial's design, status, planned outcomes, and absence of an outcome analysis; it does not provide evidence for or against a treatment effect.
13. Why This Trial Matters Statistically
This trial matters statistically because it illustrates the difference between a registered research question and an answer supported by data. The record defines an intended randomized comparison and a primary endpoint, but also states that the study ended early and outcome data were not collected. Without observations, an analysis cannot estimate the treatment contrast or quantify its uncertainty.
Randomization and masking are important design features, but they do not by themselves produce evidence of an effect. Their value is realized through collected outcomes and a suitable analysis. In this record, the planned design remains visible while the outcome comparison is absent. Keeping those two details distinct prevents a design description from being mistaken for a finding.
The record also shows why registered secondary outcomes and planned methods need to be distinguished from completed analyses. Listing outcomes does not establish that they were measured, and naming distributional checks does not demonstrate that those checks were performed. The registry records 0 analyses, so there are no findings across the listed outcome set to interpret or compare.
Finally, planning calculations and safety counts answer different questions from an efficacy analysis. Planning assumptions describe conditions for a sample-size illustration; they do not reveal an observed treatment effect. Safety event counts describe recorded events within the stated at-risk groups, but do not establish comparative risk or precision. Clear separation of plans, observed data, analyses, and uncertainty is essential when interpreting a terminated pilot study.
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 most consequential limitation is stated directly in the registry: the study was terminated early, and data were not collected due to low enrollment. This leaves no primary-outcome estimate and no analysis of the listed secondary outcomes. There is consequently no basis in the record for quantifying treatment differences or their precision.
The registered description is limited in other ways. It does not report detailed eligibility criteria beyond the summary description, the allocation ratio, the randomization procedure, or the roles covered by quadruple masking. It does not report endpoint units, a clinically important threshold, or the full method for calculating VE/VCO2 slope change. These omissions limit appraisal of the planned procedures, though they do not change the central fact that outcome data were not collected.
The record lists secondary outcomes and describes normality assessments for VO2 change outcomes, but does not provide analysis methods for the primary comparison or results of any distributional checks. It records 0 analyses. Multiplicity handling, analysis populations, and missing-data procedures are not reported in the registry record.
The safety figures are also limited: the record reports counts and at-risk groups, but does not give event definitions, collection details, precision estimates, or comparative analyses. The small reported groups do not support broad conclusions about safety. Overall, the registry is informative about the planned study and its early termination, but it cannot provide an estimate of efficacy or a precise assessment of safety.
15. Related Tutorials
16. Related Statistical Calculators
17. Sources
- ClinicalTrials.gov: Metformin in Heart Failure Without Diabetes — NCT03331861.
- Registry record last updated 2024-07-18; status last verified 2024-07.
- Results first posted 2024-07-18.
- 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.
Continue through the Clinical Biostats statistical pathway
Use this trial's endpoint as a starting point for deeper study of effect measures, confidence intervals, and sample-size planning.
18. Record Summary
The registry identifies NCT03331861 as a phase 2 randomized, quadruple-masked, parallel study sponsored by Columbia University. It compares metformin hydrochloride with matched placebo, each described for 6 months, in a study of heart failure among people without diabetes and without evidence of insulin resistance.
Recorded enrollment is 6. The study began on 2019-02-15, its primary completion date is 2021-08-20, and its status is terminated. The primary outcome was change in VE/VCO2 slope from baseline to 6 months. The registry lists 5 outcome measures and 0 analyses, and states that data were not collected because enrollment was low.
The posted record therefore describes the intended trial and its safety counts but contains no outcome comparison or treatment-effect estimate. Its planning assumptions should not be read as observed trial results.